You are on page 1of 174

The Application of Mineral Processing Techniques to the

Scrap Recycling Industry

Scott Carl Koermer

Thesis submitted to the faculty of the


Virginia Polytechnic Institute and State University
in partial fulfillment of the requirements for the degree of

Master of Science
in
Mining and Minerals Engineering

Gerald H. Luttrell, Chair


Gregory T. Adel
Michael J. Mankosa

September 2, 2015
Blacksburg, Virginia

Keywords: Recycling, Metal, Circuit Analysis, Tree Analysis

Copyright 2015, Scott Carl Koermer


THE APPLICATION OF MINERAL PROCESSING TECHNIQUES
TO THE SCRAP RECYCLING INDUSTRY
SCOTT KOERMER

ABSTRACT

The scrap metal recycling industry is a growing industry that plays an important role in the
sustainability of a large global metal supply. Unfortunately, recycling lacks many standards, and test
procedures in place for mineral processing. These standards and practices, if used in recycling, could aid
recyclers in determining and achieving optimal separations for their plant.. New regulations for scrap
imports into China make it difficult to obtain the metal recoveries that have been achieved in the past.
In order to help scrap yards adhere to the new regulations the Eriez RCS eddy current separator system
was tested in full scale. The principles this system uses, called circuit analysis, have been used by the
mining industry for years, and can be used with any separation system. The Eriez RCS system surpassed
the requirements of the Chinese regulations, while simultaneously increasing the recovery of metals. In
order to further analyze eddy current separator circuits, tree analysis was attempted for single eddy
current separators, as well as more complex circuits mimicked using locked cycle tests. The circuits used
in the locked cycle test were a rougher-cleaner, a rougher-scavenger, and a rougher-cleaner-scavenger.
It was found that it is possible to use tree analysis to compare different eddy current separator circuits
using the same settings, however standards for this practice need to be established for it to be useful.
Using the data analysis methods developed for this particular tree analysis, the rougher-cleaner-
scavenger test had the best performance overall. This is the same result as the full scale testing done on
the Eriez RCS system, but more testing should be conducted to confirm the data analysis techniques of
calculating theoretical efficiency, recovery efficiency, and rejection efficiency.
ACKNOWLEDEMENTS
First of all, I want to thank my advisor Gerald Luttrell for choosing me to be the graduate
student to work on this project. The work on this project has been immensely rewarding, and has
changed what I do for the rest of my life. Although Jerry was not expecting an extra graduate student at
the end of the semester before I started, he always found the time to edit papers I had written, and
always found the money to send me to the recycling conferences I needed to attend to learn about
recycling and make contacts within the industry. I would not be where I am today without his help.

I would also like Eriez Magnetics for it’s funding of this project. I especially would like to thank
Tim Shuttleworth for being a mentor and a guide in the recycling industry, and Mike Mankosa for being
on my committee, going over data, and coming up with new ideas. Everyone I have met at Eriez I have
learned something from, and has been a part of this project in some way.

My last committee member, Greg Adel, I would like to thank for both being on the committee,
and teaching the graduate particulate processing class. The class greatly increased my understanding of
mathematic, and I will always remember your explanation of how mineral processing can be more
difficult than rocket science. I remembered this when recently in an interview someone told me “It’s
not rocket science,” and I replied, “I want to change that”.

I would also like to thank Bob Bratton for his involvement in my graduate education. Without
Bob’s electrical skills, it may have been impossible to get the German machine used in the Chapter 3 up
and running. It has also been a privilege to learn from him and to his intuitive knowledge of the coal
processing industry.

Also I would like to thank Jim Waddell helping with everything from mounting the feeder on the
eddy current separator to helping with the design of the dust containment structure. Chapter 3 would
not have happened without Jim.

I would like to thank everyone involved in the Minerals Engineering Department at Virginia
Tech. I would like to thank all of the professors I took undergraduate classes with, as it seems like I have
used my entire education to work on this graduate degree. I would especially like to thank everyone in
the office for answering questions, doing many currency conversions, putting up with missed flights, and
ordering the items I needed for my research.

Lastly I would like to thank the lab technicians Stephen Truban and Alexander Thomson for
helping to assay and shovel the tons of material that was required to process for Chapter 3. I would
have gotten less than half of the work done for that chapter without them.

iii
TABLE OF CONTENTS
CHAPTER 1: INTRODUCTION ............................................................................................................................ 1
1.1 SCRAP METAL RECYCLING TODAY ................................................................................................................................ 1
1.2 EDDY CURRENT SEPARATORS ........................................................................................................................................ 3
1.3 PROCESS CALCULATIONS ................................................................................................................................................ 6
1.3.1 Yield ....................................................................................................................................................................................7
1.3.2 Recovery and Rejection .............................................................................................................................................8
1.3.3 Grade ............................................................................................................................................................................... 10
1.3.4 Separation Efficency ................................................................................................................................................ 12
1.3.5 Using Yield and Recovery ...................................................................................................................................... 13
1.4 GOALS OF RESEARCH .................................................................................................................................................... 15
REFERENCES........................................................................................................................................................................... 16
CHAPTER 2: CIRCUIT ANALYSIS AND TESTING THE ERIEZ RCS SYSTEM ....................................... 17
Abstract ................................................................................................................................................................................................................... 17
2.1 INTRODUCTION ............................................................................................................................................................... 17
2.2 TECHNOLOGY DESCRIPTION ......................................................................................................................................... 20
2.3 THEORETICAL DISCUSSION ........................................................................................................................................... 21
2.4 FULL SCALE TESTING..................................................................................................................................................... 24
2.5 CONCLUSIONS.................................................................................................................................................................. 26
REFERENCES........................................................................................................................................................................... 27
CHAPTER 3: LOCKED CYCLE TESTING AND TREE ANALYSIS .............................................................. 28
Abstract ................................................................................................................................................................................................................... 28
3.1 OVERVIEW AND THEORY .............................................................................................................................................. 28
3.2 TESTING APPARATUS..................................................................................................................................................... 32
3.3 PROCEDURE ..................................................................................................................................................................... 35
3.4 RESULTS AND DISCUSSION............................................................................................................................................ 40
3.4.1 Circuit Results .............................................................................................................................................................. 40
3.4.2 Tree Analysis Results ................................................................................................................................................ 42
3.4.2.1 Overview ....................................................................................................................................................................................................... 42
3.4.2.2 Zorba Grade and Recovery ................................................................................................................................................................... 42
3.4.2.3 Aluminum Grade and Recovery ......................................................................................................................................................... 46
3.4.2.4 Red Metals Grade and Recovery ........................................................................................................................................................ 49
3.4.2.4.1 Grade and Recovery ....................................................................................................................................................................... 49
3.4.2.4.2 Specific Reds Grade and Recovery .......................................................................................................................................... 52
3.4.2.5 Separation Efficiency ............................................................................................................................................................................... 56
3.4.2.6 Rejection Efficiency and Recovery Efficiency .............................................................................................................................. 61
3.5 SUMMARY AND CONCLUSION ....................................................................................................................................... 67
REFERENCES........................................................................................................................................................................... 69
CHAPTER 4: CONCLUSIONS AND RECCOMENDATIONS......................................................................... 70
APPENDIX A: TESTING OF ERIEZ RCS SYSTEM ........................................................................................ 72
APPENDIX B: RAW DATA FOR TREE ANALYSIS TESTS ......................................................................... 79
APPENDIX C: TREE ANALYSIS DATA FOR TEST 1 ................................................................................... 84

iv
APPENDIX D: TREE ANALYSIS FOR TEST 2 ............................................................................................. 105
APPENDIX E: TREE ANALYSIS DATA FOR TEST 3 ................................................................................. 125
APPENDIX F: TREE ANALYSIS FOR TEST 4 .............................................................................................. 145

v
LIST OF FIGURES
CHAPTER 1: INTRODUCTION ................................................................................................................................................ 1
Fig 1.1: Circuit boards, all separated by hand. ..........................................................................................................1
Fig. 1.2: Zorba from an eddy current separator. Currency shown for scale. ..............................................2
Fig. 1.3: A separation by an eddy current separator (Wilson et. al. 1994). ...................................................3
Fig. 1.4: Equation for centrifugal force perpendicular to the shell surface. ................................................3
Fig 1.5: Magnetic rotor causing eddy currents to be induced in conductive particles (Rem et. al.
1997) ..............................................................................................................................................................................................4
Fig 1.6: Equation for deflection due to magnetic field (Zhang et. al. 1999). ...............................................4
Fig 1.7: Force diagram for an airborne particle and a particle about to leave the belt (Zhang et.
al. 1999). .......................................................................................................................................................................................5
Fig 1.8: A sketch of an eddy current separator also showing a splitter (Zhang et. al. 1999) ..............5
Fig 1.9: Feed sample showing split by eddy current separator..........................................................................6
Fig. 1.10: Graphical result of a yield calculation. .....................................................................................................7
Fig. 1.11: Illustration of material used to calculate recovery and rejection of zorba. ............................8
Fig. 1.12: Graphical interpretation of how recovery and rejection contribute to the feed as a
whole. .............................................................................................................................................................................................9
Figure 1.13: How the recovered material is categorized for grade calculations. .................................. 10
Fig. 1.14: Graphical interpretation of the zorba grade calculation. ............................................................ 11
Fig. 1.15: Split showing only recovery of materials is used to calculate separation efficiency. ....... 12
Fig. 1.16: Visual and graphical interpretation of the mass yield of a different feed. ............................ 13
Fig. 1.17: Visual and graphical interpretation of recovery and rejection for a different feed. ........ 14
CHAPTER 2: CIRCUIT ANALYSIS AND TESTING THE ERIEZ RCS SYSTEM ................................................................... 17
Fig. 2.1. Eddy current separator (ECS). ....................................................................................................................... 18
Fig. 2.2. Example of recovery vs. purity trade-off. .................................................................................................. 18
Fig. 2.3. Misplacement mechanisms during eddy current separation. (a) Improper Orientation, (b)
Material Interaction, and (c) Splitter Fouling. ........................................................................................................ 19
Fig. 2.4. Material flows passing through the three-stage Eriez RCS eddy current separator. ........... 20
Fig. 2.5. Materials flow for the three-stage Eriez RCS eddy current separation technology.............. 21
Fig. 2.6. Comparison of misplaced material for different layouts of multi-stage eddy current
separators. ................................................................................................................................................................................ 23
Fig. 2.7. Photograph of the full-scale Eriez RSC eddy current separator. ................................................... 24
Fig. 2.8. Comparison of separation curves for traditional single-stage and three-stage Eriez RCS
eddy current separation technology............................................................................................................................. 25
CHAPTER 3: LOCKED CYCLE TESTING AND TREE ANALYSIS......................................................................................... 28

vi
Fig. 3.1 Diagram of tree analysis................................................................................................................................... 29
Fig. 3.2 The results from Meloy’s model of tree analysis done on a difficult to separate coal (Meloy
et. al. 1998) ............................................................................................................................................................................... 30
Fig. 3.3: A single eddy current separator. ................................................................................................................. 32
Fig. 3.4: Showing the enclosure for the eddy current separator. ................................................................... 33
Fig. 3.5: Copper found in the concentrate (right), and copper found in the tailings (left) ................ 33
Fig. 3.6: Constituents of the feed tested including aluminum (top left), brass (top right), ICW
(bottom right), and circuit board (bottom left). ..................................................................................................... 34
Fig. 3.7: Splitter setting used in testing. ..................................................................................................................... 35
Fig. 3.8: Stages and samples collected for the locked cycle test used in Test 2........................................ 38
Fig. 3.9: Stages and samples collected for the locked cycle test used in Test 3........................................ 38
Fig. 3.10: Stages and samples collected for the locked cycle test used in Test 4. .................................... 39
Figure 3.11: Tailings sample assay procedure, most of the sample is fluff................................................ 39
Fig. 3.12: The recirculating load in locked cycle testing as a percentage of the feed........................... 41
Fig. 3.13: Test 1 zorba grade vs. zorba recovery. .................................................................................................. 44
Fig. 3.14: Test 2 zorba grade vs. zorba recovery ................................................................................................... 44
Fig. 3.15: Test 3 zorba grade vs. zorba recovery. .................................................................................................. 45
Fig. 3.16: Test 4 zorba grade vs. zorba recovery. .................................................................................................. 45
Fig. 3.17: Test 1 aluminum grade vs. aluminum recovery................................................................................. 47
Fig. 3.18: Test 2 aluminum grade vs. aluminum recovery................................................................................. 47
Fig. 3.19: Test 3 aluminum grade vs. aluminum recovery................................................................................. 48
Fig. 3.20: Test 4 aluminum grade vs. aluminum recovery................................................................................. 48
Fig. 3.21: Test 1 reds grade vs. reds recovery. ......................................................................................................... 50
Fig. 3.22: Test 2 reds grade vs. reds recovery. ......................................................................................................... 50
Fig. 3.23: Test 3 reds grade vs. reds recovery. ......................................................................................................... 51
Fig. 3.24: Test 4 reds grade vs. reds recovery. ......................................................................................................... 51
Fig. 3.25: Showing a general calculation and the calculation used for specific grade. ....................... 52
Fig. 3.26: Test 1 specific reds grade vs. reds recovery. ........................................................................................ 54
Fig. 3.27: Test 2 specific reds grade vs. reds recovery. ........................................................................................ 54
Fig. 3.28: Test 3 specific reds grade vs. reds recovery. ........................................................................................ 55
Fig. 3.29: Test 4 specific reds grade vs. reds recovery. ........................................................................................ 55
Fig. 3.30: Test 1 fluff rejection vs. zorba recovery, rougher only. .................................................................. 57
Fig. 3.31: Test 1 fluff rejection vs. zorba recovery, rougher-cleaner. ........................................................... 58
Fig. 3.32: Test 1 fluff rejection vs. zorba recovery, rougher-scavenger. ...................................................... 58

vii
Fig. 3.33: Test 1 fluff rejection vs. zorba recovery, rougher-cleaner-scavenger. .................................... 59
Fig. 3.34: Test 2 fluff rejection vs. zorba recovery, rcs locked cycle. ............................................................. 59
Fig. 3.35: Test 3 fluff rejection vs. zorba recovery, r-c locked cycle. ............................................................. 60
Fig. 3.36: Test 4 fluff rejection vs. zorba recovery, r-s locked cycle. .............................................................. 60
Fig. 3.37: Test 1 rejection and recovery efficiency, rougher only................................................................... 63
Fig. 3.38: Test 1 rejection and recovery efficiency, rougher-cleaner. .......................................................... 64
Fig. 3.39: Test 1 rejection and recovery efficiency, rougher-scavenger. ..................................................... 64
Fig. 3.40: Test 1 rejection and recovery efficiency, rougher-cleaner-scavenger..................................... 65
Fig. 3.41: Test 2 rejection and recovery efficiency. ............................................................................................... 65
Fig. 3.42: Test 3 rejection and recovery efficiency. ............................................................................................... 66
Fig. 3.43: Test 4 rejection and recovery efficiency. ............................................................................................... 66
CHAPTER 4: CONCLUSIONS AND RECCOMENDATIONS ................................................................................................... 70
APPENDIX A: TESTING OF ERIEZ RCS SYSTEM ................................................................................................... 72
APPENDIX B: RAW DATA FOR TREE ANALYSIS TESTS ................................................................................................... 79
APPENDIX C: TREE ANALYSIS DATA FOR TEST 1 ........................................................................................................... 84
APPENDIX D: TREE ANALYSIS FOR TEST 2 ................................................................................................................... 105
APPENDIX E: TREE ANALYSIS DATA FOR TEST 3 ........................................................................................................ 125
APPENDIX F: TREE ANALYSIS FOR TEST 4 .................................................................................................................... 145

viii
LIST OF TABLES

CHAPTER 3: LOCKED CYCLE TESTING AND TREE ANALYSIS......................................................................................... 28


Table 3.1: Examples of Tree Analysis sample labeling. ....................................................................................... 36
Table 3.2: The results of individual circuits. ............................................................................................................ 40
Table 3.3: Data for circuits plotted on the specific grade vs. recovery boundaries. .............................. 52
Table 3.4: Showing data for individual circuits associated with the rejection/recovery plots. ...... 56
Table 3.5: Shows calculations for circuits tested derived from spline fit equations. ............................ 61

ix
CHAPTER 1: INTRODUCTION

1.1 SCRAP METAL RECYCLING TODAY


Recycling is defined as the reuse of a material to make a new product. Scrap metal recycling has
been around for hundreds of years. In the beginnings of the scrap metal industry, of modern times,
entrepreneurs would sort through material in dumps or refuse piles in order to identify and separate
valuable metals that they could then resell (Minter 2013). These early days sound similar to the early
days of coal processing, except today many components of a modern scrap yard are still handpicked
from the material flow, including circuit boards (Figure 1.1), and the copper from shredded electric
motors, also known as meatballs.

Fig 1.1: Circuit boards, all separated by hand.

While some materials are still separated by hand today, most material is separated by a variety
machines used to make iron, aluminum, and stainless steel products (Vishesh et. al. 2009). After
material is shredded, it is then processed using large magnets that remove ferrous particles. The
resulting non-ferrous (NF) residue is then sized and sent to the non-ferrous plant. In the non-ferrous
plant eddy current separators separate a mixture of aluminum, copper, zinc, magnesium, lead, and brass
as a product that is termed “zorba” by the industry (Figure 1.2). The tailings from the eddy current
separator circuit still contain stainless steel, among other poor conducting metals, and some copper,
and it is sent to a circuit of sensor sorters. Sensor sorters use either optical, x-ray, or metal detection in
order to sense a piece of metal on the belt of the machine and eject it from the stream either using

1
paddles or bursts of air. In a standard American Configuration the tailings from this process are then
landfilled, however other countries, particularly countries in the EU, are required by law to landfill only a
small percentage of the car being recycled (Vishesh 2009). If this is the case, there will be further
processing to remove plastics, glass, and rock.

Fig. 1.2: Zorba from an eddy current separator. Currency shown for scale.

Zorba produced in the United States is primarily sent to China and India for separation into
different types of metal by hand, and is then sold to manufacturers as a raw material. Businesses in
these countries prefer to buy mixed scrap metals, instead of pure metals, for import, not only because
labor is cheap for the separation by hand, but in order to avoid higher taxes for a more valuable, pure
product (Minter 2013). Sometimes zorba may be processed by dense medium separation if the final
metal products are going to be used in U.S. manufacturing.

The research in this paper was only conducted with the eddy current separators. Although the
methods outlined could be used with any separation process, eddy current separators were studied
because the NF part of the scrap yard drives the profitability of the shredder (Taylor 1999). Along with
the importance of NF metals recovery, China, a major importer of American scrap, recently has enacted
laws limiting the metals purity of imported mixed metals at 98%. This law, referred to by American
industry as “Operation Green Fence”, means that scrap yards will have to create higher-grade products
for their customers, and recovering less metal by making these adjustments.

There are many difficult obstacles for a process engineer working in scrap metals to overcome.
Along with the low technology practice of handpicking materials, assaying is done by hand, is not
standardized, the quality of the assay varies from person to person. Also, business is conducted from the
results of the hand assay. Samples are taken frequently, but can often be too small due to the labor

2
intensive nature of the hand assay procedure. Terms important to processing are sometimes used
incorrectly. The terms of recovery, yield, and grade are sometimes used interchangeably by staff
working at yards. Terms such as partition curve, micropricing, and tree analysis are rarely used, and
these methods are difficult to apply to recycling. The “Green Fence” now in place makes the analysis of
a scrap yard by a process engineer more important today, than ever.

1.2 EDDY CURRENT SEPARATORS


An eddy current separator is used to separate nonferrous, conductive metals. There are a few
mechanisms of an eddy current separator that can be adjusted in order to make a better or worse
separation. The main components of an eddy current separator are the belt, a magnetic rotor at the
end of the belt, and a splitter that separates material based on trajectory. A sketch of these
components, and how the work together, is shown below in Figure 1.3, which also shows the collection
of ferrous fines. Eddy current separators that sort fines frequently have a way to collect ferrous dust
that is not separated earlier upstream by large drum magnets.

Fig. 1.3: A separation by an eddy current separator (Wilson et. al. 1994).

The belt speed of the eddy current separator influences the velocity in the x direction of
particles leaving the belt, and therefore can change the quality of the separation. When a particle
leaves the belt, its velocity due to belt speed is affected by centrifugal force (Zhang et. al. 1999). An
equation for the centrifugal force of a particle at the end of a belt is shown in Figure 1.4, where L is the
length of the particle, W is the width of the particle, T is the thickness of the particle, ρ is the density of
the particle, R is the radius of the external shell at the end of the belt, and ω is the angular velocity
of the particle.

𝐹𝑐 = 𝐿𝑊𝑇𝜌𝑅𝜔2

Fig. 1.4: Equation for centrifugal force perpendicular to the shell surface.

3
The other mechanism that affects a particles trajectory is the magnetic field created by the rotor
contained within the shell at the end of the belt of the eddy current separator. The rotor is composed of
alternating magnets as shown in Figure 1.5. The alternating magnets, when the rotor is spinning at a
high frequency, creates eddy currents within a conductive particle and cause it to be ejected from the
other material.

Fig 1.5: Magnetic rotor causing eddy currents to be induced in conductive particles (Rem et. al. 1997)

The interaction between the eddy currents in conductive particles and the magnetic field is
called the Lorentz force (Rem et. al. 1997). An eddy current is essentially a current moving through the
particle in a loop, and this current is both induced and interacts with the fluctuating magnetic field
created by the rotor spinning at a high frequency. This interaction is difficult to model, however it is
possible to approximate the force on the particle due to the magnetic field. An equation for this force is
shown in Figure 1.6., where K is a complex coefficient related to magnetic roll system design, 𝐵𝑒 is the
effective magnetic induction, 𝑓 is the oscillation frequency of the magnetic field, m is the mass of the
particle, σ is the conductivity of the particle, ρ is the density of the particle, and p is a coefficient related
to shape and orientation.

𝜎
𝐹𝑑 = 𝐾𝐵𝑒2 𝑓𝑚 𝑝
𝜌

Fig 1.6: Equation for deflection due to magnetic field (Zhang et. al. 1999).

4
Effective magnetic induction can be affected by the number of poles around the circumference
of the rotor as well as the radius of the rotor. Other forces acting on the particles include frictional
forces, drag forces, gravitational forces, and interparticle forces. A diagram showing the forces acting on
a particle while in the eddy current field is shown in Figure 1.7.

Fig 1.7: Force diagram for an airborne particle and a particle about to leave the belt (Zhang et. al. 1999).

After the particle is airborne it will fall into one of two collection bins. The material that reports
to each bin can be changed by moving a splitter. A splitter is a blade that can be adjusted for angle and
possibly moved in the x and y direction depending on the maker of the machine. The splitter allows for
separation based on trajectory of the particle. A splitter is shown in the schematic in Figure 1.8 labeled
as number 4.

Fig 1.8: A sketch of an eddy current separator also showing a splitter (Zhang et. al. 1999)

5
There are many factors that determine if a particle reports to the correct fraction while being
processed by an eddy current separator. Factors that may not be so obvious and cause misplacement
are liberation and particle size. Particles not fully liberated may not receive the full effects of the
magnetic field. Smaller conductive particles may not have enough momentum in the x direction, or be
affected enough by the magnetic field to make it over the splitter.

1.3 PROCESS CALCULATIONS


In order to track the performance of a scrap yard over time, it is important to be able to so some
basic process calculations. These calculations can also be used to justify adding new separation
equipment, which is a decision that must be made carefully in tough economic times. The calculations
illustrated below can help an operator make better decisions regarding equipment purchases and scrap
sales, as well as increasing communication regarding efficiency to employees and other scrap yards.

Figure 1.9 shows a sample that has been fed into an eddy current separator. Although the ratio
of metals to waste is above average, this hypothetical sample has been chosen to simplify calculations.
This feed sample contains twelve blocks of equal weight. The four silver blocks are pure aluminum, the
four red-orange blocks are pure copper, and the four black blocks are plastic. When separated, the
blocks shown on the red background go under the splitter of the eddy current separator and are sent to
a landfill, while the blocks on the green background go over the splitter and report to the zorba product.

Fig 1.9: Feed sample showing split by eddy current separator.

6
1.3.1 YIELD

The most common measure of separation performance is yield. Yield is simply the mass
percentage of feed that ends up in a separated product, Most commonly, yield is used to describe the
fraction that is more valuable (i.e., zorba yield). In Figures 1.9 and 1.10, zorba yield is indicated as the
area within the green background. Note that this area contains both desirable metals and unwanted
misplaced plastic. Because the blocks in the diagram are equal in weight, the yield for both fractions is
50%. A yield describing the material that will be sold as zorba is calculated by:

𝑀𝑎𝑠𝑠 𝑜𝑓 𝐶𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑒
𝑌𝑖𝑒𝑙𝑑 (%) = 100 ×
𝑀𝑎𝑠𝑠 𝑜𝑓 𝐹𝑒𝑒𝑑

6 𝐵𝑙𝑜𝑐𝑘𝑠
𝑌𝑖𝑒𝑙𝑑 (%) = 100 × = 50%
12 𝐵𝑙𝑜𝑐𝑘𝑠

Mass
Yield
50%

Fig. 1.10: Graphical result of a yield calculation.

A sudden change in yield can be caused by changes in feed composition or by an equipment


malfunction that impacts separation performance.

7
1.3.2 RECOVERY AND REJECTION

Fig. 1.11: Illustration of material used to calculate recovery and rejection of zorba.

Recovery and rejection calculations are used to show the distribution of a specific component
present in the feed between the separated products. These parameters are important for assessing the
overall performance of the process. For process engineers, both recovery and rejection measurements
are better indicators of separation performance than yield or the metals content in the waste of a
process. For example, the valuable metal blocks that have been “recovered” are shown in the green
area on Figure 1.11 and 1.12. Zorba recovery, which is the percentage of recovered metals out of the
total metals in the feed, is calculated as:

𝑀𝑎𝑠𝑠 𝑜𝑓 𝑧𝑜𝑟𝑏𝑎 𝑚𝑒𝑡𝑎𝑙𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑒


𝑍𝑜𝑟𝑏𝑎 𝑅𝑒𝑐𝑜𝑣𝑒𝑟𝑦 (%) = 100 ×
𝑀𝑎𝑠𝑠 𝑜𝑓 𝑧𝑜𝑟𝑏𝑎 𝑚𝑒𝑡𝑎𝑙𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑓𝑒𝑒𝑑

5 𝐵𝑙𝑜𝑐𝑘𝑠
𝑍𝑜𝑟𝑏𝑎 𝑅𝑒𝑐𝑜𝑣𝑒𝑟𝑦 = 100 × = 63%
8 𝐵𝑙𝑜𝑐𝑘𝑠

Thus, for every 100 tons of metals sent to the separator, 63 tons were recovered as valuable
zorba product. The remaining 38 tons were lost to the waste stream.

8
The 38 tons of metals that were lost can be described by rejection. Rejection is the opposite of
recovery. Rejection is normally used to quantify the percentage of a specific type of material that
reported to the waste stream in a separation process. In the current example, the blocks in the red area
of Figure 1.11 represent those that reported to the waste stream. As such, zorba metals rejection is
calculated as:

𝑀𝑎𝑠𝑠 𝑜𝑓 𝑧𝑜𝑟𝑏𝑎 𝑚𝑒𝑡𝑎𝑙𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑡𝑎𝑖𝑙𝑖𝑛𝑔𝑠


𝑍𝑜𝑟𝑏𝑎 𝑅𝑒𝑗𝑒𝑐𝑡𝑖𝑜𝑛 = 100% − 𝑅𝑒𝑐𝑜𝑣𝑒𝑟𝑦 = 100 ×
𝑀𝑎𝑠𝑠 𝑜𝑓 𝑧𝑜𝑟𝑏𝑎 𝑚𝑒𝑡𝑎𝑙𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑓𝑒𝑒𝑑
3 𝐵𝑙𝑜𝑐𝑘𝑠
𝑍𝑜𝑟𝑏𝑎 𝑅𝑒𝑗𝑒𝑐𝑡𝑖𝑜𝑛 = 100% − 62% = 100 × = 38%
8 𝐵𝑙𝑜𝑐𝑘𝑠

Metals
Rejection
38% Metals
Recovery
63%

Fig. 1.12: Graphical interpretation of how recovery and rejection contribute to the feed as a whole.

In this case the metals rejection is shown as the red area in both the pie chart and the material diagram.
Once again the shaded area, which represents plastic, is not used in the calculation for rejection or
recovery of metals. However, rejection can also be used to describe the amount of plastic that ends up
in the zorba, and would help to describe the amount of contamination in the product. Therefore
designing a process with a higher plastic rejection could mean a higher grade zorba product. Recovery
and rejection are also useful for determining metal losses per hour, which then can be converted into
revenue losses using current market prices.

9
1.3.3 GRADE

Grade describes the amount of a specific constituent present in a scrap sample, i.e., product
purity. The grade of a pile of scrap is important for both sales and processing calculations. The blocks in
the non-shaded concentrate region shown in Figure 1.13 are used to calculate the grade of the zorba.
Of the non-shaded blocks, 50 percent of them are aluminum, making the grade of the zorba produced in
this process 50% aluminum. If indicated, the grade can also describe the overall metals content, which
is the sum of the grade of aluminum and the grade of copper. Aluminum grade is calculated as shown
below:

𝑀𝑎𝑠𝑠 𝑜𝑓 𝑎𝑙𝑢𝑚𝑖𝑛𝑢𝑚 𝑖𝑛 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑒


𝐴𝑙𝑢𝑚𝑖𝑛𝑢𝑚 𝐺𝑟𝑎𝑑𝑒 = 100 ×
𝑇𝑜𝑡𝑎𝑙 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑒

3 𝐵𝑙𝑜𝑐𝑘𝑠 𝑜𝑓 𝐴𝑙
𝐴𝑙𝑢𝑚𝑖𝑛𝑢𝑚 𝐺𝑟𝑎𝑑𝑒 = 100 × = 50%
3 𝐵𝑙𝑜𝑐𝑘𝑠 𝑜𝑓 𝐴𝑙 + 2 𝐵𝑙𝑜𝑐𝑘𝑠 𝑜𝑓 𝐶𝑢 + 1 𝐵𝑙𝑜𝑐𝑘 𝑜𝑓 𝑃𝑙𝑎𝑠𝑡𝑖𝑐

Figure 1.13: How the recovered material is categorized for grade calculations.

10
Plastic
17%
Al
50%
Cu
33%

Fig. 1.14: Graphical interpretation of the zorba grade calculation.

Grade can also describe the metal content in the feed and waste streams from a separator.
Therefore, waste grade is also an important parameter for monitoring metal losses.

11
1.3.4 SEPARATION EFFICENCY

Separation efficiency is a technical metric that shows the overall effectiveness of the separation.
For the current example, separation efficiency would typically be reported as the recovery of metals in
the zorba product minus the recovery of plastic in the metals product. Separation efficiency can be
calculated as:

𝑆𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛 𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 = 𝑍𝑜𝑟𝑏𝑎 𝑀𝑒𝑡𝑎𝑙𝑠 𝑅𝑒𝑐𝑜𝑣𝑒𝑟𝑦 − 𝑃𝑙𝑎𝑠𝑡𝑖𝑐 𝑅𝑒𝑐𝑜𝑣𝑒𝑟𝑦

𝑆𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛 𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 = 63% − 25% = 38%

Fig. 1.15: Split showing only recovery of materials is used to calculate separation efficiency.

Figure 1.15 shows both the recovery and rejection of plastic and metals, and the dotted lines in
the figure are color coded to the separation efficiency equation. Only the recovery of plastic and the
recovery of metals are used in the calculation. Rejection, which describes the distribution of material to
the shaded area, is not used.

12
1.3.5 USING YIELD AND RECOVERY

While yield’s simple calculation does not involve measuring grade, it is not the best indicator of
separation performance. Since recovery and rejection describe the percent of metals that reach the
customer, it is a much better indication of how well the separation is performing and if improvements
can be made. In the example shown below in Figure 1.16, the feed composition and how the material
splits changes; yet the yield stays the same, making yield a poor indicator of separation performance.

Mass
Yield
50%

Fig. 1.16: Visual and graphical interpretation of the mass yield of a different feed.

As shown in the pie chart in Figure 1.16, the yield for the changed feed is still 50%. However, in
Figure 1.17 it is apparent that the metals recovery has dropped from 63% to 57%, and the overall
separation efficiency has dropped from 38% to 17%. Copper was the only metal where the recovery
dropped: dropping from 50% to 33%. With copper being the more valuable metal, this drop in recovery
would create a problem for the processor. If only yield is measured and grade measurements are not
completed consistently, the processor may not be aware in the drop in performance. This drop in
recovery can also be an indicator of a change in shape of the material, possibly indicating poor shredder
performance, a problem with sizing, or the separators themselves.

13
Metals
Rejection
43%
Metals
Recovery
57%

Fig. 1.17: Visual and graphical interpretation of recovery and rejection for a different feed.

A 5% loss in metals recovery can have a large effect over time. If a scrap yard runs only 1/4 of a
ton per hour of copper through eddy current separators, and the copper recovery drops by 17%, that is
the equivalent of losing 476,000 lb of copper every year if the separators are in operation 16 hours a day
for 50 weeks out of the year. Knowing these numbers, and how to use them, lets the processor know
where to look to get every last penny out of their scrap.

14
1.4 GOALS OF RESEARCH
The goal of this research and this thesis paper is to help start an improvement in the way
recycling is done today.

This could be in terms of:

 Product design for future separation.


 Improvements in software used to monitor scrap yards.
 More standard practices, including standards in assaying and testing.
 An understanding of actual metal losses.
 An understanding of what testing could be done next.

,among many other possibilities.

15
REFERENCES
1. Minter, Adam. Junkyard Planet: Travels in the Billion-dollar Trash Trade. New York: Bloomsbury,
2013.

2. Recycling Today Staff, ISRI Convention: China’s Green Fence. Recycling Today, April 23, 2013.

3. Rem, P. C., P. A. Leest, and A. J. Van den Akker. "A model for eddy current
separation." International journal of mineral processing 49, no. 3 (1997): 193-200.

4. Taylor, B., Providing the profits. Recycling Today, August 1999.

5. Vishesh Kumar, John W. Sutherland, Development and assessment of strategies to ensure


economic sustainability of the U.S. automotive recovery infrastructure, Resources, Conservation
and Recycling, Volume 53, Issue 8, June 2009, Pages 470-477, ISSN 0921-3449,
http://dx.doi.org/10.1016/j.resconrec.2009.03.012.

6. Wilson, R.J., T.J. Veasey, D.M. Squires, The application of mineral processing techniques for the
recovery of metal from post-consumer wastes, Minerals Engineering, Volume 7, Issue 8, August
1994, Pages 975-984, ISSN 0892-6875, http://dx.doi.org/10.1016/0892-6875(94)90027-2.

7. Zhang, S., E. Forssberg, B. Arvidson, W. Moss, (1999). Separation mechanisms and criteria of a
rotating eddy-current separator operation.Resources, conservation and recycling, 25(3), 215-
232.

16
CHAPTER 2: CIRCUIT ANALYSIS AND
TESTING THE ERIEZ RCS SYSTEM

ABSTRACT

Eddy current separators are commonly used in the recycling industry to recover non-ferrous metals
from nonmetallic waste residues. Recently, the push for higher grade products from market consumers
has created an incentive for designers to improve the separation efficiency of these machines. In
response to this challenge, a new multi-stage eddy current separator has been developed based on a
fundamental concept developed in process engineering. Production runs conducted with a full-scale
industrial machine showed that this new technology could readily generate zorba products with purity
levels exceeding 98% metals while maintaining mass recoveries of metals in excess of 95%. As such, this
new technology makes it possible for plant operators to meet increasingly stringent purity targets
desired by customers, while also improving profitability by increasing the amounts of metals recovered
from wastes.

2.1 INTRODUCTION
Production statistics compiled by the Institute of Scrap Recycling Industries (ISRI) indicate that in
2014 the scrap recycling industry in the United States annually processes more than 113 million metric
tons of scrap, including 74 million metric tons of ferrous scrap and 8 million metric tons of nonferrous
scrap. Recyclable metals present in the nonferrous scrap include many high-value “zorba” metals such
as aluminum, brass, copper, nickel and zinc. Since these non-ferrous metals do not respond to magnetic
separations, they are often recovered using dense medium (float-sink) processes that require
wetting/dewatering of the scrap feedstock. Scrap yard operators may also elect to sell their Zorba
products to foreign contractors whose lower pay scales permit labor-intensive manual sorting
(handpicking) to be economically viable. However, this approach usually does not provide the scrap yard
operator with the largest financial return on their saleable products.

An attractive option for sorting nonferrous metals from waste is eddy current separation (Zhang
et al., 1999). As shown in Figure 2.1, an ECS machine consists of a rotor encompassed by a shell made of
magnets with alternating magnetic poles, which is contained inside a conveyor belt drum. When rotated
at high speed, the rotor induces eddy currents within nonmagnetic, conductive metals. The induced
field deflects nonferrous metals up and away from the conveyor drum, while non-metallic materials
remain unaffected by the field and simply drop from the end of the conveyor. During operation,
nonferrous metals and nonmetallic waste is fed onto the conveyor belt in a thin layer. The belt then
passes over the rotor, where the induced

17
1.

Fig. 2.1. Eddy current separator (ECS). Fig. 2.2. Example of recovery vs. purity trade-off.

eddy currents cause zorba metals to become repelled from the rotor. The repulsion causes the
trajectory of the non-ferrous metal to be different than the nonmetallic contaminants. The two streams
of material are separated by an adjustable splitter into metals and non-metals products. The splitter
position can either be (i) moved outward to improve the purity of the metals product at the expense of
mass recovery or (ii) moved inward to improve the mass recovery at the expense of product purity. The
trade-off between product recovery and product quality typically results in the creating of a recovery-
grade curve such as that shown in Figure 2.2.

A large financial incentive exists to shift the separation curve towards the right-upper corner of
the recovery-grade plot, i.e., towards maximum recovery and maximum purity. This shift is possible by
reducing the amount of material that is inadvertently misplaced during an eddy current separation. The
misplacement of metals and nonmetals is caused by a wide variety of reasons including the improper
orientation of objects entering the eddy current field (Figure 2.3a), interferences created by material
contacts (Figure 2.3b), and fouling of splitters as a result of material hang-ups (Figure 2.3c). Whatever
the cause, misplacement shifts the recovery-grade curve away from the optimum operating region and
adversely impacts product recovery and/or quality.

18
Fig. 2.3. Misplacement mechanisms during eddy current separation. (a) Improper Orientation, (b) Material
Interaction, and (c) Splitter Fouling.

In light of the misplacement problem, process engineers from Eriez and Virginia Tech have
applied advanced concepts originally developed in the minerals processing industry to develop a next-
generation eddy current separator that can provide a higher quality metal product at a higher recovery.
This new innovation, called the Eriez RCS eddy current separator, incorporates multiple stages of
cleaning (to improve quality) and scavenging (to improve recovery) into a single compact unit. The new
design is capable of overcoming the longstanding dilemma created by the trade-off between metals
recovery and metals grade that is unavoidable with a single-stage machine.

19
2.2 TECHNOLOGY DESCRIPTION
It has long been known that multi-stage processing systems can be used to improve the
recovery or purity of recycled products. Multi-stage layouts that involve reprocessing of the valuable
product to improve purity and reduce contamination are typically called “cleaning circuits.” Likewise,
layouts that involve reprocessing of the waste product to improve the recovery of valuable components
are called “scavenging circuits.” Cleaning circuits improve product purity at the expense of recovery,
while scavenging circuits improve recovery at the expense of product purity. However, if properly
configured, combinations of cleaning and scavenging circuits can be used together to simultaneously
improve both product recovery and purity.

Fig. 2.4. Material flows passing through the three-stage Eriez RCS eddy current separator.

Figure 2.4 provides a simplified schematic of material flows through the new Eriez RCS eddy
current separator. During operation, mixed metal/waste scrap is initially passed through a “rougher”
stage to sort the feed into metallic and non-metallic products. The metals product from the rougher
stage is then passed to a “cleaner” stage of separation to eliminate any nonmetallic waste that may
have been misplaced into the saleable product. Likewise, the nonmetallic product from the rougher
stage is passed through a “scavenger” stage of separation to recover any metals that may have been
accidentally misplaced into the waste stream. These two stages of cleaning and scavenging ensure that
little misplaced material is left in either the metals or waste products. The most novel feature of the
Eriez RCS machine, however, is the manner in which the off-spec streams (i.e., waste stream from the
cleaner unit and off-spec metals stream from the scavenger unit) are further processed. These two off-
spec streams are recycled back to the feed of the rougher stage of the Eriez RCS (Rougher-Cleaner-
Scavenger) machine. As discussed below, the forced recirculation of off-spec material creates a self-

20
regulating loop of cleaning and scavenging that is superior in terms of separation efficiency compared to
any other layout of process units.

2.3 THEORETICAL DISCUSSION


A process engineering tool known as linear circuit analysis (Meloy, 1983) can be used to explain
why the new Eriez RCS eddy current separator offers such a high level of separation performance. This
powerful tool has been used to design multi-stage circuits for size classification (Honaker et al., 2007),
density concentration (Luttrell et al., 1998; McKeon and Luttrell, 2005) and magnetic separation (Luttrell
et al., 2004). According to the circuit analysis concept, the tonnage of a particular piece of material
reporting to the metal concentrate (C) can be calculated by multiplying the tonnage of the material in
the feed (F) by a separation probability (P). Mathematically, this relationship can be defined as:

C
C  P( F ) or P [1]
F
For the case of perfect separation, valuable metallic objects would have Pmetal=1 and non-
metallic wastes would have Pwaste=0. Unfortunately, real-world separators suffer from misplacement
since some portion of the metals may be lost to waste (Pmetal<1) and some waste may be inadvertently
passed into the metals concentrate (Pwaste>0). To reduce this misplacement, the Eriez RCS technology
reprocesses both the metals concentrate and the non-metals waste as shown in Figure 2.5. The waste
rejected from the final metals concentrate and the metals recovered from the final non-metals waste
are recycled back to the feed for reprocessing. According to linear circuit analysis, the concentrate (C)
produced by the Eriez RCS technology can be calculated as:

C  F ' Pr Pc [2]

Fig. 2.5. Materials flow for the three-stage Eriez RCS eddy current separation technology.

Pr, Pc and Ps are the separation probabilities for the rougher, cleaner and scavenger stages, respectively.
Since two streams are recycled back to the feed (F), the new internal feed (F’) passed to the eddy
current separators is given by:

21
F  F 'F ' Pr (1  Pc )  F ' (1  Pr ) Ps ` [3]

F  F '[1  Pr (1  Pc )  (1  Pr ) Ps ] ` [4]

F
F' ` [5]
1  Pr (1  Pc )  (1  Pr ) Ps

By combining Equations [2] and [5], the concentrate-to-feed (C/F) ratio for the Eriez ECS technology can
be calculated as:

 F 
C  Pr Pc [6]
1  Pr (1  Pc )  (1  Pr ) Ps 

C Pr Pc
 ` [7]
F 1  Pr (1  Pc )  (1  Pr ) Ps

If it is assumed that the separation probability (P) is identical for all three separators (i.e., P=Pr=Pc=Ps),
then Equation [7] simplifies to:

C P2
 ` [8]
F 1  2P  2P 2

If a single-stage eddy current separator recovered 90% of the metals (P=0.9) and 10% of the waste
(P=0.1), then the three-stage Eriez RCS technology would be expected to give:

C P2 0.92
Metals in Zorba :    0.988 [9]
F 1  2P  2 P 2 1  2(0.9)  2(0.92 )

C P2 0.12
Waste in Zorba :    0.0122 [10]
F 1  2P  2 P 2 1  2(0.1)  2(0.12 )

According to this analysis, the Eriez RCS system would reduce the amount of waste recovered in
the zorba concentrate from 10% down to only 1.2% (P=0.0122) while simultaneously increasing the
amount of metals recovered in the zorba concentrate from 90% to 98.8% (P=0.988). Therefore, the use
of three stages of separation coupled with the recirculation of off-spec material back to the feed, the
Eriez RCS technology is capable of producing (i) high-purity low-impurity zorba concentrates that are
attractive to customers and (ii) higher recoveries and lower losses of metals that can increase the overall
profitability of recycling operations.

22
Fig. 2.6. Comparison of misplaced material for different layouts of multi-stage eddy current separators.

Using linear circuit analysis, the recovery of metals and non-metals (waste) in the zorba
concentrate can be estimated for several other configurations of two- and three-stage eddy current
systems. The comparative results, which are plotted in Figure 2.6, indicate that no other layout is
capable of achieving the high-purity and high-recovery performance of the three-stage Eriez RCS
technology. Interestingly, the analysis indicates that a three-stage layout without recirculation (Layout 5)
would be expected to perform no better than a single-stage unit when the off-spec products are passed
forward and recombined with their corresponding metals or waste products. This type of poorly

23
designed layout represents three times the capital investment with no benefit in terms of separation
performance.

2.4 FULL SCALE TESTING


In order to fully demonstrate the improved performance capabilities of a multi-stage eddy
current separator, a commercial-scale prototype (Eriez RSC Eddy Current Separator) was designed,
constructed and tested. The innovative prototype, which is shown in Figure 2.7, made it possible to
compare the performance of an integrated “optimal” three-stage machine to that of a single-stage
separator. The Eriez RSC machine was tested using a 5 ton batch of automotive shredder residue having
an upper particle size of 32 mm (1¼ inches) and containing 7 to 10% zorba mixed with waste fluff. The
RSC machine processed almost 10,000 lb of material during testing. Each set of circuit tests were run
with varying feed tonnage rates and splitter settings to provide a wide range of operating conditions and
performance values. During testing, representative samples of both waste fluff and concentrate zorba
were collected at the same time intervals during the duration of the test until the mass of zorba
collected satisfied a minimum sampling requirement of 34 kg (75 pounds) or greater. The zorba
concentrate was then hand-sorted to determine the product purity (i.e., percentage of metals in the
total zorba product), while the nonmetallic waste was slowly passed over a laboratory-scale eddy
current separator to determine the unrecovered metals content. From this data, the zorba purity and
mass recovery was calculated.

Fig. 2.7. Photograph of the full-scale Eriez RSC eddy current separator.

24
Figure 2.8 shows recovery-grade curves that compare the performance data collected from the
single-stage separator and the three-stage separator. As expected, the separation obtained using the
three-stage Eriez RSC machine with recirculation was far superior to that achieved using only the single-
stage machine. The single-stage machine, which failed to achieve 98% metals purity, was able to obtain
a grade of 93% metals at a recovery of just under 85% by weight during the “best” test run. In contrast,
the three-stage Eriez RSC machine easily achieved the target of 98% purity while obtaining metals
recoveries of 97-99% by weight. In fact, this high-efficiency machine had to be pushed to abnormal
operating conditions for the metal recovery in the zorba product to fall below 90%.

Fig. 2.8. Comparison of separation curves for traditional single-stage and three-stage Eriez RCS eddy current
separation technology.

From a business perspective, the three-stage cleaner-scavenger circuit would be expected to


offer an attractive payback based on the improved mass yield of metals product. For example, if 13 tons
per hour of automotive shredder residue containing 9% zorba metals were fed into a conventional eddy
current separator, the single-stage machine has the potential to recover 1.17 tons per hour of pure

25
metals. Using a 98% product grade, the experimental recovery-grade curves shown in Figures 6 and 7
indicate that the single eddy current separator would only recover about 45% of these metals. In
contrast, the high-efficiency Eriez RSC machine that incorporates an optimal cleaner-scavenger layout
with recirculated secondary streams would recover about 97% of the metals. Therefore, in these two
scenarios, the single eddy current separator would generate 0.537 tons per hour of metals compared to
1.16 tons per hour for the three-stage system. Using these values, the Eriez RSC machine would increase
zorba output by 116% over the single-stage unit. Assuming a fair market zorba price of $1.50 per kg
($0.68 per lb), and assuming that both systems are run for 40 hours a week for 50 weeks a year, the
116% increase in output for the Eriez RSC machine equates to an additional $1.7 million in annual
revenue.

2.5 CONCLUSIONS
A high-efficiency separator, known as the Eriez RCS eddy current separator, has been developed
to assist the metals recycling industry in attaining improved levels of separation efficiency. To
demonstrate the improved performance of this new technology, a full-scale machine was constructed
and tested using a sample of automotive shredder residue containing non-ferrous metals and non-
metallic wastes. Full-scale testing of this new machine showed tremendous increases both in the mass
recovery of metals and in the purity of the zorba product. The data obtained with the advanced three-
stage process indicated that purity levels exceeding 98% metals in the zorba product could be readily
obtained at mass recoveries in excess of 95%. The large improvement in separation efficiency was in
very good agreement with theoretical predictions obtained using a mathematical engineering tool called
linear circuit analysis (Meloy, 1983). The multi-stage machine advocated by this process engineering tool
makes it possible for recycling plant operators to meet increasingly stringent purity targets desired by
customers, while also improving operator profitability by increasing the amounts of metals recovered
from recycled wastes.

26
REFERENCES

1. Honaker, R.Q., Boaten, F., Shin, D., Luttrell, G.H., 2007. “Evaluation of High-Efficiency Classifying
Cyclone Circuits,” Proceedings, SME Annual Meeting and Exhibit, February 25-28, 2007, Denver,
Colorado, Preprint 07-143, 7 pp.

2. Luttrell, G.H., Kohmuench, J.N., Stanley, F.L. and Trump, G.D., 1998. “Improving Spiral
Performance Using Circuit Analysis,” Minerals and Metallurgical Processing, Vol. 5, No. 4,
November 1998, pp. 16-21.

3. Luttrell, G.H., Kohmuench, J.N., and Mankosa, M.J., 2004. “Optimization of Magnetic Separator
Circuit Configurations,” SME Annual Meeting, Denver, Colorado, February 24-25, 2004, Preprint
04-90, 5 pp.

4. Meloy, T.P., 1983. “Analysis and Optimization of Mineral Processing and Coal-Cleaning Circuits –
Circuit Analysis,” International Journal of Mineral Processing, Vol. 10, p. 61-72.

5. McKeon, T., Luttrell, G.H., 2005. “Application of Linear Circuit Analysis in Gravity Separator
Circuit Design” Society for Mining, Metallurgy, and Exploration, 2005. 6 pp.

6. Zhang, S., Forssberg, B.A., Moss, W., 1999. “Separation Mechanisms and Criteria of a Rotating
Eddy-Current Separator Operation” Resources, Conservation and Recycling, Vol. 25, No. 3-4, pp.
215-232.

27
CHAPTER 3: LOCKED CYCLE TESTING AND
TREE ANALYSIS
ABSTRACT

Tree analysis and release analysis have been used to analyze the separation of coal from ash in froth
flotation circuits for many years. An experiment was developed to determine if the same type of
analysis could be used for the separation of zorba metals from automotive shredder residue. For the
testing only one separator was available, and a locked cycle testing procedure had to be developed for
complex circuits. It was determined that it is possible to compare different circuits processing the same
feed using this kind of analysis, however it is important to develop a standardized procedure for
assaying and the tree analysis itself for future testing.

3.1 OVERVIEW AND THEORY


In order to gauge how well a separator is performing compared to its optimal operating point it
is important to obtain data that helps a process engineer build both grade vs. recovery curves and
rejection vs. recovery curves. When it is known how a machine performs compared to the optimum
separation for that machine, it is possible to use that data to justify capital expenditures to add units to
a separation circuit, or to rearrange circuits.

Tree analysis is a testing method, used in coal flotation, which is used to create a grade recovery
boundary for a certain feed coal under certain conditions. In coal flotation testing standards for tree
analysis are not always clear, and can sometimes produce different results. According to the paper ‘An
Evaluation of the Flotation Response of Coals’ (Pratten et. al. 1989), several parameters are pertinent to
the measurement of flotation response, in regards to release and tree analysis. These parameters are
listed below:

1. “A cumulative yield/cumulative ash curve should be able to be constructed from the results of
the procedure”
2. “The yield/ash locus must represent the limit of flotation separation.”
3. The procedure should be based on a flotation separation instead of another type of separation,
such as density.
4. “The results of this procedure should depend solely on the coal and the identity of the flotation
reagents used”
5. “The procedure should be used for the entire flotation feed…narrow size fractions can lead to
erroneous conclusions…”
6. The procedure should be simple, be able to be performed routinely, and must be repeatable.

28
Although those parameters are used to describe methods of release analysis and tree analysis in
flotation, they can be translated into parameters that can be used for tree analysis using eddy current
separators:

1. A cumulative metals yield, cumulative grade curve should be able to be constructed from the
results of the procedure.
2. The yield/grade locus must represent the limit of separation.
3. The procedure should be based on the same variables used in eddy current separation.
4. The results of this procedure should depend solely on the contents of the feed and the machine
settings.
5. The procedure should be used for the entire feed, not narrow size distributions or individual
types of metal.
6. The procedure should be simple, be able to be performed routinely, and must be repeatable.

Tree analysis was the chosen analysis method for this experiment due to its simplicity. In tree
flotation a coal feed is floated under certain conditions in a laboratory, and the resulting concentrate
and tailings subjected to re-floatation in successive scavenger and cleaner stages (Pratten 1989). This
causes the process circuit diagram to branch out, as shown below in Figure 3.1. This diagram of
branches is why the procedure is known as tree analysis.

Fig. 3.1 Diagram of tree analysis.

The resulting samples at the bottom of the diagram are assayed and weighed. Then the samples are
ordered from the highest grade to the lowest grade. Using this order the cumulative grade of clean coal
and the cumulative mass of clean coal are calculated, from the highest grade sample to the lowest grade
sample. The cumulative grade and cumulative mass numbers are used as the points for the grade vs.
recovery boundary.

The number of levels in a tree analysis has the potential to change the resulting grade vs. recovery
boundary, if standards are not put in place. In the paper “Flotation tree analysis – reexamined” it is

29
brought to the readers attention that the number of branches used in the tree analysis, could potentially
change the shape of the grade vs. recovery boundary (Meloy et. al 1998). In a model used for the paper
Meloy showed how, with a completely liberated coal, the boundary can shift based on the number of
levels in the tree. The results of the model are shown below in Figure 3.2. The labels under the curve
that read 4, 20, and 100 refer to the amount of levels of the tree analysis.

Fig. 3.2 The results from Meloy’s model of tree analysis done on a difficult to separate coal (Meloy et. al. 1998)

Although the model used in “Flotation tree analysis - reexamined” lacks complexity it does make
a valid point apparent, which is very important for release analysis and the way it would potentially be
performed in the recycling industry. In scrap recycling assays are conducted by hand, and a particle that
is 51% metal and 49% plastic would most likely be counted as a particle of 100% metal. Essentially
assays in the scrap metal recycling industry frequently assume there is complete liberation of metal.
Therefore, for a tree analysis of the performance of eddy current separators, it is important to put a
constraint on the number of levels used in the procedure.

The paper “Froth flotation: preparation of a laboratory standard” discusses the importance of
coming up with and international standard laboratory procedure for both release analysis, and tree
analysis (Brown, Hall 1999). The testing in this paper used a standard of re-floating samples until there
was no more concentrate produced, or the mass of either the concentrate or tailings of a flotation was
less than 5% of the mass of solids in the slurry (Brown, Hall 1999). Although many standards were

30
outlined in Brown’s paper, this is the most applicable standard to tree analysis for eddy current
separators. Paralllels can be drawn to the standard use of frothers, collectors, and flocculants to the use
of specific splitter, and belt speed settings, but before testing these settings were not apparent, and
would be different for different machines. Using the known information about tree analysis in coal
flotation, an experiment was designed and performed.

This experiment uses locked cycle tests in order to mimic complex circuits. This is necessary in
laboratory environments, because it may be difficult or impossible to obtain and arrange the number of
separators required for these complex circuits. The locked cycle test allows the technician to mimic
circuits that use recirculating loads, when only one machine is available for testing.

31
3.2 TESTING APPARATUS
For this test, a small eddy current separator was used, similar to the eddy current separator
pictured in Figure 3.3.

Fig. 3.3: A single eddy current separator.

Collection bins were placed under the eddy current separator in order to collect tailings samples
and concentrate samples from material that was run. An Eriez 6” vibratory feeder was used along with
an Eriez control unit that varied the output of the vibratory feeder by percent of output. A hopper
found in the lab was modified so that it could feed the material to the vibratory feeder. The hopper and
feeder were then mounted on the eddy current separator, and a chute was made to drop the material
on a 6 inch by 6 inch area on the belt. There was a lip on the belt to remove ferrous particles that may
get stuck in the rotor area, so this was the best option.

After the hopper and feeder were mounted on the eddy current separator, and the entire unit
was enclosed in a structure built for dust containment. The dust containment structure is shown below
in Figure 3.4, and did not interfere with the testing. The top of the hopper can be shown enclosed by
the wooden box, pictured at top left of Figure 3.4. The Eriez control unit for the vibratory feeder is
hanging from a stud with two orange wires connected to the bottom.

32
Fig. 3.4: Showing the enclosure for the eddy current separator.

The material itself was sized to -1 1/4 inch, and zorba metals were mixed with automotive
shredder residue in order to produce a non-ferrous residue for testing. The size distribution of the
material is fairly wide. Pictured in Figure 3.5 are the copper pieces that were found in the concentrate,
at the right, and copper particles found in the tailings at the left. Because there was not bottom size all
particles under 1.25 inches that came out of the shredder would be in this fraction.

Fig. 3.5: Copper found in the concentrate (right), and copper found in the tailings (left)

33
Other constituents of interest of the feed are aluminum, brass, circuit boards, and insulated
copper wire (ICW) pictured in Figure 3.6. The feed also included fluff, which is a generic mixture of
plastics, fabrics, dirt, wood, foam, and glass.

Fig. 3.6: Constituents of the feed tested including aluminum (top left), brass (top right), ICW (bottom right), and
circuit board (bottom left).

Other apparatus used included a forklift with a barrel lifter attachment, shovels, and buckets.

34
3.3 PROCEDURE
First, zorba metals were mixed with automotive shredder residue (ASR) in order to create a
residue containing nonferrous metals. The metals were mixed with the ASR so that the resultant
residue contained about 10% aluminum and 2% red metals, which are brass, bronze, and bare copper,
by weight. Batches of 200 lb were mixed at a time, which was about the capacity of the hopper and the
large bin in place to collect fluff. All materials were weighed in a plastic drum, sealed, and tumbled five
times using a barrel lifter.

Before this material was run, the settings for the eddy current separator needed to be chosen.
These settings would stay constant for the entire test. Settings were chosen that were expected to
produce a good separation. The settings were chosen, in this case, based on the operator’s experience
of these parameters effect the separation, in both real world testing and simulation software. The belt
speed was chosen to be constant at 2.0 m/s. Belt speed changes the horizontal velocity and travel of all
particles. It may be necessary to increase belt speed in order to recover finer aluminum and copper
particles, which are not effected as much by the eddy current field, however a greater belt speed can
produce a lower grade, concentrate. The setting of 2.0 m/s is somewhat fast, but not at the fastest
setting of 2.5 m/s allowed by the control unit.

The splitter setting used is shown in Figure 3.7. The belt and rotor are just right of what is in the
illustration. This setting was chosen after running some material at a low feed rate with this splitter
setting. It appeared to create a mostly clean product. The federate was arbitrarily chosen to be 60% on
the vibratory feeder control box. This was found to be a low federate, about 1.15 tons per hour per foot
of belt width, and allowed the federate to be increased for locked cycle testing. Because of the chute
design, the federate is a maximum value, the true federate is unknown and is slightly lower due to
material spreading out on the belt.

Fig. 3.7: Splitter setting used in testing.

35
For the first test 3 200lb barrels were run in order to collect about 70 lb of material of
concentrate needed for assaying. A limitation for this experiment was the amount of material available
for testing. Ideally one would want at least a 70 lb sample for every product of the tree analysis, but this
was not possible due to the amount of material available for testing and time constraints. The products
of the first separation were then split following a path similar to the one shown in Figure 3.1. For this
tree analysis, a material was not run through the eddy current separator more than six times. By the
sixth time the ASR was processed, only a small mass of metal was removed by the eddy current
separator. The pure concentrate, and the pure tailings were processed six times, however if it was
determined that a sample towards the middle of the tree was much too small of a sample size, or
further processing may cause a substantial loss of material, the branch was stopped.

Samples were labeled according to the following convention:

 For material that originally reported to the zorba concentrate, the label started with a C.
 For material that originally reported to the fluff tailings, the label started with a T.
 After the second level of processing a C or T would be the second letter, or digit, depending on
weather the material reported to the concentrate or tailings.
 After the nth level of processing a C or T would be the nth letter, or digit, depending on weather
the material reported to the concentrate or tailings.
 The number of digits in the label indicates the number of times the material has been processed
by the eddy current separator.

Some examples are shown in Table 3.1:

Label Description

CCC,CCC The material has been processed 6 times and


reported to the concentrate every time.

TC,CTC The material has been processed 5 times and


has reported to the tailings, then the
concentrate, the concentrate, the tailings and
then the concentrate, in that order.

TTC The material has been processed 3 times. The


first and second runs it reported to the
tailings. After the third level of processing it
reported to the concentrate. The sample was
not run further due to small sample size.

Table 3.1: Examples of Tree Analysis sample labeling.

Using this convention it is easy to determine, how many times the material has been processed,
and its location on the tree.

36
Test 1 was conducted in order to obtain results from rougher, rougher-cleaner, and rougher-
scavenger circuits. Tests 2, 3, and 4 employed rougher-cleaner-scavenger, rougher-cleaner, and
rougher-scavenger tests with recirculating loads. The tree analysis from these tests were conducted in
the same manner as Test 1, except the split indicated by “R” in Figure 3.1 was the result of a locked cycle
test.

For Test 2 the locked cycle test mimicked a rougher-scavenger-cleaner circuit. In order to
conduct the locked cycle test 200 lb batches of nonferrous residue were run, and samples were
collected as shown in Figure 3.8. All settings on the eddy current separator were kept constant for all
stages. The same settings for the eddy current separator were used for every stage of every test.

For the first run of the locked cycle test middlings, from the first level scavenger (MT) and the
first level cleaner (MC), do not exist until the material is processed and can be ignored as part of the
feed. After the first run the middling’s produced from the nonferrous residue are combined and
weighed. Then, another standard 200 lb of nonferrous residue were mixed, but the middling’s were also
added in. The locked cycle circuit shown in Figure 3.8 is completed again. The setting on the control
box is adjusted in proportion to the increase in feed mass due to the added middlings. The concept
behind changing the feed rate is, if the circuit had to output a constant amount of material per hour, but
now also had to process middlings, the added middlings may reduce the performance of the circuit. This
forces the feed rate of the circuit to become an independent variable instead of a dependent variable.

Both the T and C samples are discarded until the mass of the middlings does not change much
for two or more runs. This means that the recirculating load is stable, and therefore the locked cycle
test is mimicking the intended RCS circuit. Material is run through the locked cycle circuit until about 70
lb of concentrate is collected.

For Test 3, a locked cycle test that mimicked a rougher-cleaner circuit was used. Once again
nonferrous residue was processed until the mass of the middlings stabilized, then samples were
collected at this point. The locked cycle diagram for Test is depicted in Figure 3.9. Test 4 conducted in
the same manner as Test 2 and Test 3, except a rougher-scavenger circuit was used. The locked cycle
diagram for Test 4 is shown in Figure 3.10.

After enough sample was collected for all locked cycle tests, the products of the test were run
through a tree analysis with the same procedure as Test 1.

All samples were assayed after collection. Due to the learning curve of assaying, some of the
procedures changed from test to test, but the methods were kept constant for individual tests. Samples
were assayed for aluminum, aluminum wire, bare copper, brass/bronze, circuit boards, insulated copper
wire, and fluff. Figure 3.11 shows a tailings assay being conducted. Any ferrous or stainless steel
particles found were included in the fluff. The material that went to tailings every time was coned and
quartered in order to obtain a smaller, representative sample estimated to be 75 lb at the time of coning
and quartering.

37
Fig. 3.8: Stages and samples collected for the locked cycle test used in Test 2.

Fig. 3.9: Stages and samples collected for the locked cycle test used in Test 3.

38
Fig. 3.10: Stages and samples collected for the locked cycle test used in Test 4.

Figure 3.11: Tailings sample assay procedure, most of the sample is fluff.

39
3.4 RESULTS AND DISCUSSION

3.4.1 CIRCUIT RESULTS

The results from specific circuits are shown in Table 3.2. From the tree analysis from Test 1, it
was possible to calculate the results of more complex circuits that did not require a recirculating load.

Table 3.2: The results of individual circuits.

Test Zorba Zorba Al Al Red Metals Red Metals


Number Description Grade Recovery Grade Recovery Grade Recovery

Test 1 Rougher Only 94.05 86.09 77.80 97.37 16.25 55.37

Test 1 Rougher- Cleaner 98.07 84.58 81.30 95.87 16.77 53.85

Test 1 Rougher-Scavenger 93.42 87.97 75.87 97.69 17.55 61.51

Rougher-Cleaner-
Test 1 Scavenger 97.29 86.46 79.15 96.19 18.13 59.99

Test 2 R-C-S Locked Cycle 98.94 91.37 86.83 96.91 12.11 64.83

Test 3 R-C Locked Cycle 98.73 88.24 85.79 94.01 12.94 62.73

Test 4 R-S Locked Cycle 95.00 86.17 83.40 94.16 11.60 53.52

These results do not show much variation from test to test. There are a few reasons for this,
including the small recirculating load that was used in the locked cycle tests. Figure 3.12 shows the mass
of the middlings, used as a recirculating load, as a percent of the new feed added to the locked cycle
test. The middlings weights for Test 2 were not recorded for each run, however the mass eventually
stabilized at about 1% of the feed or 2 lb. This small mass makes the recirculating streams have a
minimal impact on the performance, compared to the large impact shown in the Eriez RCS system.

40
Run Number vs. Percent of Feed
Recirculating
Middlings Produced (% of Feed)

0.8
0.7
0.6
0.5
0.4
Test 3
0.3
0.2 Test 4
0.1
0
0 1 2 3 4 5 6
Locked Cycle Run Number

Fig. 3.12: The recirculating load in locked cycle testing as a percentage of the feed.

Another possibility that could explain unexpected variations in the data is that the feed changed
over time. The material was run many times, and eventually some of the particles can break down,
effectively changing the size distribution.

The performance of locked cycle tests can also change the size distribution in the feed, which
would have an impact on the next test. It is very likely that this had an effect on Test 4. Table 3.2 shows
that the zorba recovery for Test 4, which used a rougher scavenger, is less than in Test 3, which used a
rougher cleaner. This is opposite than what would be expected. Before Test 4 was run, another test
was run that had a recirculating load upwards of 28% of the feed. This was achieved by using a splitter
position that was much closer to the belt than all other tests. A particle’s size has a large impact on the
distance it is thrown by an eddy current separator (Zhang et. al. 1998). Moving the splitter in, allowed
more fine particles to be recovered and concentrated in the middlings, which were mixed in with new
feed. This could have changed the size distribution of metals in the feed. When the splitter was moved
back out for Test 4 these particles had a much lower probability of ending up in the concentrate, but still
counted as a loss when conducting an assay.

41
3.4.2 TREE ANALYSIS RESULTS

3.4.2.1 OVERVIEW
This section covers the results obtained through tree analysis. In order to generate the
operating boundary samples were first ordered from the highest to the lowest grade. The definition of
grade may change between plots depending on the material of interest, for example copper grade is
different than zorba grade. Then the mass distribution of all of the constituents was calculated. From
there cumulative grade, recovery, and rejection could be calculated for each sample in the order of the
descending grade. All data, calculated numbers, tree diagrams, and larger versions of the plots can be
viewed in Appendices 2, 3, 4, and 5.

3.4.2.2 ZORBA GRADE AND RECOVERY


Using the assayed samples from Tree analysis, it was possible to generate Grade vs. Recovery
boundaries. The plots in Figures 3.13 to 3.16 show grade vs. recovery boundaries for all tests in regards
to zorba. The circuits indicated by distinct points on the plots represent the data in Table 3.2.

In Figure 3.13 all of the circuits that could be derived from Test 1 are shown. The rougher
scavenger has a slightly higher recovery than the rougher, but a worse grade. The rougher cleaner is
operating in a different direction with a slightly lower recovery, but a higher grade. These are both as
expected. The rougher-cleaner-scavenger circuit is operating with both a higher grade and a higher
recovery, but neither value is quite as high as the highest value achievable with a two unit circuit. The
rougher-cleaner-scavenger circuit would theoretically be expected to produce the same results as just
the rougher.

It is clear that none of the points in Figure 3.13 are operating at the boundary. This means that
with the same feed and machine settings used, there could potentially be an improvement in efficiency
if different circuits were used.

Figure 3.14 shows the results of the rougher-cleaner-scavenger locked cycle test and tree
analysis. This point is operating very close to the elbow of the boundary generated from tree analysis. It
is not right on the elbow, but it is still closer than any of the points in Figure 3.13.

Figure 3.15 shows the results of the rougher-cleaner locked cycle test and tree analysis. It
appears that the point showing the results of the first separation is lower in relation to the boundary on
the y-axis. This illustrates a lower recovery than what is optimal at that grade. The difference in grade
from the point to the boundary appears to be about the same as it is in Test 2.

In Figure 3.16 the results from the rougher-scavenger locked cycle test are plotted. The grade
for the locked cycle test has been pushed further away from the boundary, but the distance from the
point to the boundary on the y-axis appears to be less than it was for the rougher-cleaner locked cycle
test.

42
The boundary point, shown on the plots in Figures 3.13-3.16, that has the highest recovery and
the lowest grade, should be noted. The sample was used to create that point had a lower zorba grade
than any other sample, besides the material that always went to the tailings. During tree analysis the
material that ended up in the pure tailings sample TTT,TTT went under the splitter as a reject 6 or 7
times. All of the material that ends up in this sample could potentially be deemed unrecoverable. In a
scrap yard an eddy current circuit with legs that were 5 or 6 units long may be not feasible economically.
Knowing this, the leftmost point shown on the grade/recovery boundary is the maximum recovery
possible with the machine settings and feed used in that test if the metals that ended up in TTT,TTT are
considered unrecoverable.

It should be noted that Test 4 had the lowest possible recovery, while Test 2 had the highest
possible recovery on the boundaries.

43
Test 1 Zorba Grade vs. Zorba Recovery
100

95 Tree Analysis
Zorba Recovery (%)

90 Test 1 Rougher

85 Rougher Cleaner
Scavenger
Rougher Scavenger
80
Rougher Cleaner
75

70
70 75 80 85 90 95 100
Zorba Grade (%)

Fig. 3.13: Test 1 zorba grade vs. zorba recovery.

Test 2 Zorba Grade vs. Zorba Recovery


100

95 Tree
Analysis

90
Recovery (%)

Test 2 RCS
85 Locked
Cycle

80

75

70
70 75 80 85 90 95 100
Grade (%)

Fig. 3.14: Test 2 zorba grade vs. zorba recovery

44
Test 3 Zorba Grade vs. Zorba Recovery
100

95
Tree Analysis
90
Recovery (%)

85
Rougher Cleaner
80 Locked Cycle

75

70
70 75 80 85 90 95 100
Grade (%)

Fig. 3.15: Test 3 zorba grade vs. zorba recovery.

Test 4 Zorba Grade vs. Zorba Recovery


100

95
Tree
Analysis
90
Recovery (%)

85 Rougher
Scavenger
Locked Cycle
80

75

70
70 75 80 85 90 95 100
Grade (%)

Fig. 3.16: Test 4 zorba grade vs. zorba recovery.

45
3.4.2.3 ALUMINUM GRADE AND RECOVERY
Grade/recovery boundaries were generated in the same manner as in section 3.4.2.2, except
aluminum grade was used instead of zorba grade. All data, calculated numbers, and larger plots can be
seen in Appendicies 3-6.

For Test 1, shown in Figure 3.17, both the rougher scavenger and the rougher had fairly high
recoveries of aluminum. Although the rougher cleaner scavenger outperformed the rougher overall on
zorba, the rougher has a higher recovery of aluminum than the rougher cleaner scavenger. The rougher
cleaner has the highest grade.

In Test 2, shown in Figure 3.18, the RCS locked cycle test once again performed very close to the
elbow of the boundary. The RCS point in Test 2 may be able to improve slightly in regards to grade.

Test 3, shown in Figure 3.19, shows the rougher cleaner locked cycle test. It is operating very
close to the most efficient point on the curve.

Test 4, shown in Figure 3.20, shows the rougher scavenger locked cycle test. It appears this
point is close to the optimal aluminum recovery, but it could improve in grade.

It should be noted that the maximum grade for each plot is between 80% and 90%. This is
because any zorba material, such as bare copper wire, is considered a contaminant, but still has a higher
probability it will end up in the concentrate than fluff. Using these machine settings, it would be
impossible to obtain a perfect separation between red metals and aluminum, no matter how many
machines were used.

46
Test 1 Aluminum Grade vs. Aluminum
Recovery
100
Tree Analysis
95
Aluminum Recovery (%)

Test 1 Rougher
90
Rougher Cleaner
85 Scavenger
Rougher
80 Scavenger
Rougher Cleaner
75

70
70 75 80 85 90 95 100
Aluminum Grade (%)

Fig. 3.17: Test 1 aluminum grade vs. aluminum recovery.

Test 2 Aluminum Grade vs. Aluminum


Recovery

100

95 Tree
Analysis
Recovery (%)

90

85
Test 2 RCS
80 Locked
Cycle
75

70
70 75 80 85 90 95 100
Grade (%)

Fig. 3.18: Test 2 aluminum grade vs. aluminum recovery.

47
Test 3 Aluminum Grade vs. Aluminum
Recovery
100

95 Tree
Analysis

90
Recovery (%)

85 Rougher
Cleaner
80 Locked Cycle

75

70
70 75 80 85 90 95 100
Grade (%)

Fig. 3.19: Test 3 aluminum grade vs. aluminum recovery

Test 4 Aluminum Grade vs. Aluminum


Recovery
100

95 Tree
Analysis
90
Recovery (%)

85
Rougher
Scavenger
80 Locked Cycle

75

70
70 75 80 85 90 95 100
Grade (%)

Fig. 3.20: Test 4 aluminum grade vs. aluminum recovery.

48
3.4.2.4 RED METALS GRADE AND RECOVERY

3.4.2.4.1 GRADE AND RECOVERY


For all tests red metals grade vs. red metals recovery plots were generated along with the
associated grade/recovery boundary. As shown in Figures 3.21-3.24, these boundaries are in an
unconventional shape. The same methods were used to generate these plots that were used to
generate zorba grade vs. recovery plots and aluminum grade vs. recovery plots.

When these plots were generated, aluminum was considered a contaminant for the calculation
of red metals grade. This gives them the unconventional shape. These plots show that the maximum
possible grade of red metals for a circuit with these settings on each machine is between 35% and 65%.

49
Test 1 Reds Grade vs. Reds Recovery
100
90 Tree Analysis
80
70 Test 1 Rougher
Recovery (%)

60
50 Rougher Cleaner
40 Scavenger

30
Rougher
20 Scavenger
10
Rougher Cleaner
0
0 20 40 60 80 100
Grade (%)

Fig. 3.21: Test 1 reds grade vs. reds recovery.

Test 2 Reds Grade vs. Reds Recovery


100
90
80
70
Tree Analysis
Recovery (%)

60
50
Test 2 RCS
40 Locked Cycle
30
20
10
0
0 20 40 60 80 100
Grade (%)

Fig. 3.22: Test 2 reds grade vs. reds recovery.

50
Test 3 Reds Grade vs. Reds Recovery
100
90
80 Tree Analysis
70
Recovery (%)

60
50 Rougher Cleaner
Locked Cycle
40
30
20
10
0
0 20 40 60 80 100
Grade (%)

Fig. 3.23: Test 3 reds grade vs. reds recovery.

Test 4 Reds Grade vs. Reds Recovery


100
90
80 Tree
Analysis
70
Recovery (%)

60
50
Rougher
40 Scavenger
Locked Cycle
30
20
10
0
0 20 40 60 80 100

Grade (%)

Fig. 3.24: Test 4 reds grade vs. reds recovery.

51
3.4.2.4.2 SPECIFIC REDS GRADE AND RECOVERY
The boundary for red metals grade vs. red metals recovery is unconventional. A calculation
called specific grade was generated in order to better describe this process. The calculated values for
the overall separations in each test are shown in Table 3.3.

Table 3.3: Data for circuits plotted on the specific grade vs. recovery boundaries.

Test Number Description Specific Reds Grade Red Metals Recovery

Test 1 Rougher Only 73.18 55.37

Test 1 Rougher- Cleaner 89.68 53.85

Test 1 Rougher-Scavenger 72.72 61.51

Test 1 Rougher-Cleaner-Scavenger 86.98 59.99

Test 2 R-C-S Locked Cycle 91.94 64.83

Test 3 R-C Locked Cycle 91.06 62.73

Test 4 R-S Locked Cycle 69.90 53.52

Specific grade is a term, which was created to analyze this data in a way that made visual
comparisons of plots easier. Specific grade only takes into account the specified material and the waste
material. In this case aluminum does not fall into either of those categories and aluminum content does
not change the specific grade of a sample. Specific grade is calculated as shown in Figure 3.25.

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙 𝑥
𝑔𝑟𝑎𝑑𝑒𝑥 =
𝑚𝑎𝑠𝑠 𝑜𝑓 𝑥 + 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑤𝑎𝑠𝑡𝑒

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑟𝑒𝑑 𝑚𝑒𝑡𝑎𝑙𝑠


𝑔𝑟𝑎𝑑𝑒𝑟𝑒𝑑𝑠 =
𝑚𝑎𝑠𝑠 𝑜𝑓 𝑟𝑒𝑑 𝑚𝑒𝑡𝑎𝑙𝑠 + 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑙𝑢𝑓𝑓

Fig. 3.25: Showing a general calculation and the calculation used for specific grade.

Specific grade allows the grade to be more indicative of the ratio of the material of interest to
the material not expected to be separated by an eddy current separator.

Test 1 (Figure 3.26) shows that the rougher cleaner scavenger did pretty well on the separation
of red metals. The point is close to the boundary, and has a slightly lower grade than the rougher

52
cleaner, and a slightly lower recovery than the rougher scavenger. The rougher scavenger recovered
more red metals than the rougher at a slightly lower grade, while the rougher cleaner had a much higher
specific grade of red metals for a small red metal loss. According to this boundary, the maximum red
metal recovery is 64.7% At this point it seems that the grade recovery boundary is dependent on
machine settings and feed content. Points are shifted within the boundary by using different circuits.

Test 2 (Figure 3.27) shows that the red metals recovery is close to the boundary at that specific
grade. A higher specific grade does not necessarily mean a higher red metals to aluminum ratio,
however If another scavenger was added to the circuit it should increase the red metals recovery.
Aluminum recovery was almost on the boundary for Test 2, therefore not much more aluminum could
be recovered. For that reason an increase in red metals recovery should mean an increase in the ratio
of red metals to aluminum. For Test 2, the maximum red metals recovery is 69.8%.

For Test 3 (Figure 3.28) the rougher cleaner locked cycle test had a slightly larger gap between
the boundary and the point on the y-axis than there was in Test 2. For this test the maximum red metals
recovery achievable is 69%.

In Test 4 (Figure 3.29) there is a dramatic drop in specific reds grade, both in absolute value and
relative to the specific grade/recovery boundary. This makes sense due to the nature of the scavenger
circuit to create a product with a lower grade. The maximum potential recovery according to the tree
analysis is 57.9%. This is much lower than what was possible in other tests. This may be due to the
possibility of the change in the size distribution of the feed, or the material breaking down over time.

53
Test 1 Specific Reds Grade vs. Reds Recovery
100
Tree
Analysis
90
Rougher
80
Recovery (%)

Rougher
70
Cleaner
Scavenger
60 Rougher
Scavenger
50
Rougher
40 Cleaner

30
30 40 50 60 70 80 90 100
Grade (%)

Fig. 3.26: Test 1 specific reds grade vs. reds recovery.

Test 2 Specific Reds Grade vs. Reds Recovery

100

90 Tree Analysis
80
Recovery (%)

70

60
Rougher
50 Cleaner
Scavenger
40 Locked Cycle

30
30 40 50 60 70 80 90 100
Grade (%)

Fig. 3.27: Test 2 specific reds grade vs. reds recovery.

54
Test 3 Specific Reds Grade vs. Reds Recovery
100

90
Tree
80 Analysis
Recovery (%)

70

60
Rougher
50 Cleaner
Locked Cycle
40

30
30 40 50 60 70 80 90 100
Grade (%)

Fig. 3.28: Test 3 specific reds grade vs. reds recovery.

Test 4 Specific Reds Grade vs. Reds Recovery


100

90
Tree
80 Analysis
Recovery (%)

70

60 Rougher
Scavenger
Locked Cycle
50

40

30
30 40 50 60 70 80 90 100
Grade (%)

Fig. 3.29: Test 4 specific reds grade vs. reds recovery.

55
3.4.2.5 SEPARATION EFFICIENCY
In order to account for changes in feed, Zorba Recovery vs. Fluff Rejection plots were created
from the data obtained through tree analysis. The data for the individual points is shown in Table 3.4.
Theoretical efficency is a term used to describe the separation efficency of a circuit as a percent of the
maximum separation efficency shown on the fluff rejection/zorba recovery boundary. Figures 3.30-3.36
show the rejection/recovery boundary for each circuit. Each of these plots has lines that indicate
separation efficencies including the separation efficency of the circuit, and the maximum separation
efficency on the boundary.

Table 3.4: Showing data for individual circuits associated with the rejection/recovery plots.

Test Fluff Zorba Separation Theoretical


Number Description Rejection Recovery Efficiency Efficiency

Test 1 Rougher Only 99.14 86.09 85.22 96.93

Test 1 Rougher- Cleaner 99.74 84.58 84.32 95.90

Test 1 Rougher-Scavenger 99.02 87.97 86.99 98.93

Rougher-Cleaner-
Test 1 Scavenger 99.62 86.46 86.08 97.90

Test 2 R-C-S Locked Cycle 99.87 91.37 91.24 98.56

Test 3 R-C Locked Cycle 99.83 88.24 88.08 96.63

Test 4 R-S Locked Cycle 99.35 86.17 85.52 98.34

For Test 1 the maximum separation efficency on the rejection/recovery boundary is about
88.2%. The rougher, shown in Figure 3.30 had a separation efficency fo 85.22, which was 96.93% of the
maximum separation efficency on the boundary. This is a higher separation efficency than the rougher
cleaner, shown in Figure 3.31, which had a separation efficency of 84.32. Because the grades of all of
the circuits are fairly high, the fluff rejection is also high. This makes makes the changes in recovery
between circuits have a larger impact on the separation efficency. This is shown in the rougher-
scavenger, in Figure 3.32 which has the highest recovery and separation efficency at 87.97 and 86.99
respectively. The rougher-scavenger also has the highest theoretical efficency at 98.93. The rougher-
scavenger-cleaner, shown in Figure 3.33, had a theoretical efficency of 95.9%, one percent less than the
rougher-scavenger, and had the third highest separation efficency in Test 1 at 86.08.

The RCS locked cycle test in Test 2, shown in Figure 3.34, has the highest separation efficency of
any of the tests. The RCS locked cycle test also has the second highest theoretical efficency at 98.56.
There were some slight assaying differences between Test 1 and 2, due to the learning curve of assaying
the material by hand, and this in part could have caused this test to not have the highest theoretical
efficency.

56
Test 3, shown in Figure 3.35, the rougher-cleaner locked cycle test, had the lowest theoretical
efficency out of any of the locked cycle tests. This makes sense due to the high fluff rejection for all
tests.

For Test 4, shown in Figure 3.36, the separation efficency of the rougher-scavenger test is the
lowest out of all of the locked cycle tests, but the theoretical efficency is the second highest of the
locked cycle tests at 98.34%.

Using rejection, recovery, separation efficency, and theoretical efficency can be a better
indicator of which circuits have the best performance, than grade and recovery alone. The potential
differences in feed and assay accuracy could be minimized, and it may make more sense to compare a
test point to the boundary curve instead of an absolute value on the plot.

Test 1 R: Fluff Rejection vs. Zorba Recovery


100
Tree
Analysis
95 Test 1
Zorba Recovery (%)

Rougher
SE = 80
90

SE = 90
85
SE = 88.2

80 SE = 85.2

75
75 80 85 90 95 100
Fluff Rejection (%)

Fig. 3.30: Test 1 fluff rejection vs. zorba recovery, rougher only.

57
Test 1 R-C: Fluff Rejection vs. Zorba Recovery
100
Tree Analysis

95 Test 1 Rougher
Cleaner
Zorba Recovery (%)

SE = 80
90

SE = 90
85
SE = 88.2

80
SE = 84.3

75
75 80 85 90 95 100
Fluff Rejection (%)

Fig. 3.31: Test 1 fluff rejection vs. zorba recovery, rougher-cleaner.

Test 1 R-S: Fluff Rejection vs. Zorba


Recovery
Tree
100
Analysis

95 Test 1
Rougher
Zorba Recovery (%)

Scavenger
90 SE = 80

85 SE = 90

80
SE = 87

75
75 80 85 90 95 100 SE= 88.2
Fluff Rejection (%)

Fig. 3.32: Test 1 fluff rejection vs. zorba recovery, rougher-scavenger.

58
Test 1 R-C-S: Fluff Rejection vs. Zorba
Recovery
Tree Analysis
100

95 Test 1 Rougher
Cleaner
Zorba Recovery (%)

Scavenger
90 SE = 80

85 SE = 90

80 SE = 88.2

75
SE = 86
75 80 85 90 95 100
Fluff Rejection (%)

Fig. 3.33: Test 1 fluff rejection vs. zorba recovery, rougher-cleaner-scavenger.

Test 2: Fluff Rejection vs. Zorba Recovery


100
Tree
Analysis
95
SE = 80
Recovery (%)

90 SE = 90

85 SE = 92.5

SE = 91.25
80

RCS Locked
75 Cycle
75 80 85 90 95 100
Rejection (%)

Fig. 3.34: Test 2 fluff rejection vs. zorba recovery, rcs locked cycle.

59
Test 3: Fluff Rejection vs. Zorba Recovery
100
Tree
Analysis
95
R-C Locked
Cycle
Recovery (%)

90 SE = 80

SE = 90
85
SE = 88

80 SE = 91.2

75
75 80 85 90 95 100
Rejection (%)

Fig. 3.35: Test 3 fluff rejection vs. zorba recovery, r-c locked cycle.

Test 4: Fluff Rejection vs. Zorba Recovery


100
Tree
Analysis
95 R-S Locked
Cycle
SE = 50
Recovery (%)

90

SE = 80
85
SE = 90

80 SE = 87

75 SE = 85.5
75 80 85 90 95 100
Rejection (%)

Fig. 3.36: Test 4 fluff rejection vs. zorba recovery, r-s locked cycle.

60
3.4.2.6 REJECTION EFFICIENCY AND RECOVERY EFFICIENCY
After theoretical efficiency was found for all of the circuits, the rejection/recovery boundary was
fit with a cubic spline using the average line slope method. This gave an equation that could be used to
find the difference in recovery from one point to the optimal recovery on the boundary; given both
points have the same rejection. The same could be done for rejection, given both points have the same
recovery. In Table 3.5 the values for “Efficient Recovery” are the maximum possible recovery, according
to the rejection/recovery boundary, at the rejection that resulted from the circuit used in the test.
Values for “Efficient Rejection” follow the same rules as efficient recovery, except that these values are
calculated at the recovery resulting from the circuit tested. It should be noted that it is impossible to
achieve both an efficient recovery and an efficient rejection at the same time, unless the circuit is
operating on the boundary. Plots showing the results of circuits tested in Tests 1 through 4 are shown in
Figures 3.37-3.43 along with the efficient rejection and recovery for each operating point.

Table 3.5: Shows calculations for circuits tested derived from spline fit equations.

Test Efficient Efficient Rejection Recovery


Number Description Rejection Recovery Efficiency Efficiency

Test 1 Rougher Only 99.82 89.07 99.32 96.65

Test 1 Rougher- Cleaner 99.89 87.07 99.85 95.58

Test 1 Rougher-Scavenger 99.57 89.12 99.45 98.70

Rougher-Cleaner-
Test 1 Scavenger 99.78 87.76 99.83 97.52

Test 2 R-C-S Locked Cycle 99.92 91.82 99.95 99.52

Test 3 R-C Locked Cycle 99.90 89.98 99.94 98.07

Test 4 R-S Locked Cycle 99.77 87.53 99.59 98.45

Figure 3.37 shows that for the rougher only in Test 1, there could be an increase in fluff rejection
at the same recovery. For the same rejection that was obtained in the rougher only test there could also
be an increase in recovery. These differences are shown by the distances from the blue point, indicating
the results of the test, and the two other points lying on the rejection/recovery boundary.

Figure 3.38 shows a decrease in the distance between the efficient rejection on the boundary
and the rejection that resulted from the rougher-cleaner test. This would be expected from a rougher
cleaner circuit. Comparing the numbers for recovery efficiency for the rougher cleaner and rougher ,
the rougher has a higher recovery efficiency as expected. The rejection efficiency and recovery
efficiency describe how the point for a test has moved relative to the rejection/recovery boundary.

61
In Figure 3.39 the rejection efficiency and recovery efficiency of the rougher-scavenger test is
shown, and referencing Table 3.5 it can be seen that the rejection efficiency has decreased and the
recovery efficiency has increased relative to the rougher cleaner test, as one would expect. However,
the rejection efficiency is greater than the rougher only when the rougher would make a better grade
material. This is because as recovery gets higher, the maximum rejection gets lower. If a certain
recovery number needed to be met, then the rougher may be deemed not a feasible option to use.

In Figure 3.40 the point indicating the rougher-cleaner-scavenger test appears to have a fairly
high recovery efficiency and rejection efficiency. Looking at Table 3.5, the rougher-cleaner-scavenger
has the second highest recovery efficiency and rejection efficiency for Test 1.

Figure 3.41 shows the results of Test 2. The rougher-cleaner-scavenger locked cycle test had the
highest rejection efficiency and recovery efficiency of any of the tests. This test was the closest to the
“elbow” of the curve. If this test was producing the desired fluff rejection, then it would be difficult to
justify spending money on another cleaner unit.

The RCS locked cycle test also had the highest separation efficiency and the second highest
theoretical efficiency. According to the tree analysis, a higher separation efficiency can be reached with
the same feed and machine settings. If a higher recovery was desired, and rejection could drop a little,
it would make sense to try a locked cycle test with a rougher, one cleaner, and two scavengers to see
where the test point shifted along the boundary. If the desired recovery and rejection was reached,
then it may be possible to justify capital expenditure for another unit.

Figure 3.42 shows a highly efficient rejection for the rougher cleaner locked cycle test. For this
test, the recovery decreased, and due to the sharpness of the possible separation shown on the
boundary, the recovery efficiency also decreased. It is very unlikely that a test will fall exactly on the
boundary. Therefore, a test with a low recovery, even with a rejection falling next to the boundary, still
has a chance of having a low rejection efficiency.

Figure 3.43 shows that the rougher-scavenger locked cycle test had a higher recovery efficiency,
but a lower rejection efficiency than Test 3. Looking back at the test of the Eriez RCS system in Chapter
2, it is expected that a rougher scavenger with a recirculating load would have a lower recovery and a
higher rejection than the same circuit without a recirculating load. Looking at only recovery efficiency
and rejection efficiency, the rougher scavenger with the recirculating load has a higher rejection and
lower recovery, relative to the bounds of the curve, than the rougher scavenger without the
recirculating load.

Using recovery efficiency and rejection efficiency may be a way to compare tests that use
different feed, machine settings, or assay procedures. Throughout this experiment the assay procedure
changed due to the nature of doing an assay by hand. It took time for the technician to learn how to
identify metals, and to do so accurately. Also after learning more about the material, more categories of
material were assayed. There may be some general human error in the assays including error due to
fatigue. The different quality in assays and feed could be considered noise in the data. Rejection
efficiency and recovery efficiency were created in an attempt to smooth out this noise and compare the
test points to what is possible instead of on an absolute scale.

When comparing tests where different tree analysis is done, it may make sense to use this
method due to a change in feed or machine settings. Tests that use the same tree analysis can be

62
compared on just grade, recovery, and rejection. When comparing circuits between tests comparing
rejection efficiency and recovery efficiency, gives the same changes one would expect through circuit
analysis. For example the rougher-scavenger has the highest recovery efficiency of any of the tests. The
rougher-cleaner-scavenger locked cycle test also has the best combination of both efficiencies. Since
this method was used just for this paper, more testing is be required in order to see if it is a valid tool for
analysis.

Test 1: Fluff Rejection vs. Zorba Recovery


100

98
Tree
Analysis
96

94 Test 1
Rougher
Recovery (%)

92
Efficent
90 Rejection

88 Efficent
Recovery
86

84

82
82 87 92 97
Rejection (%)

Fig. 3.37: Test 1 rejection and recovery efficiency, rougher only.

63
Test 1: Fluff Rejection vs. Zorba Recovery
100

98 Tree
Analysis
96
Rougher
94
Recovery (%)

Cleaner
92
Efficent
90 Rejection
88 Efficent
Recovery
86

84

82
82 87 92 97
Rejection (%)

Fig. 3.38: Test 1 rejection and recovery efficiency, rougher-cleaner.

Test 1: Fluff Rejection vs. Zorba Recovery


100

98 Tree
Analysis
96
Rougher
94 Scavenger
Recovery (%)

92
Efficent
90 Rejection

88 Efficent
Recovery
86

84

82
82 87 92 97
Rejection (%)

Fig. 3.39: Test 1 rejection and recovery efficiency, rougher-scavenger.

64
Test 1: Fluff Rejection vs. Zorba Recovery
100
Tree Analysis
98

96
Rougher
94 Cleaner
Recovery (%)

Scavenger
92
Efficent
90 Rejection

88 Efficent
Recovery
86

84

82
82 87 92 97
Rejection (%)

Fig. 3.40: Test 1 rejection and recovery efficiency, rougher-cleaner-scavenger.

Test 2: Fluff Rejection vs. Zorba Recovery


100

98 Tree
Analysis
96
RCS
94
Locked
Recovery (%)

92 Cycle
Efficent
90 Rejection

88 Efficent
Recovery
86

84

82
82 87 92 97
Rejection (%)

Fig. 3.41: Test 2 rejection and recovery efficiency.

65
Test 3: Fluff Rejection vs. Zorba Recovery
100

98 Tree
Analysis
96
Rougher
94
Cleaner
Recovery (%)

92 Locked Cycle
Efficent
90 Rejection
88
Efficent
86 Recovery
84

82
82 87 92 97
Rejection (%)

Fig. 3.42: Test 3 rejection and recovery efficiency.

Test 4: Fluff Rejection vs. Zorba Recovery


100

98 Tree Analysis

96

94 Rougher
Recovery (%)

Scavenger
92 Locked Cycle

90 Efficent
Rejection
88

86 Efficent
Recovery
84

82
82 87 92 97
Rejection (%)

Fig. 3.43: Test 4 rejection and recovery efficiency.

66
3.5 SUMMARY AND CONCLUSION
Tree analysis and locked cycles tests are extremely valuable tools for judging circuit
performance. If only one machine is available for testing by a manufacturer, it is possible to see how
that machine may perform as part of a larger, complex circuit.

When looking at just the grade and recovery of the circuits used in the tests, not all of them
perform as expected in relation to one another. In general the circuits that used a locked cycle test
performed better than their counterparts that did not use a recirculating load. For example the rougher
cleaner locked cycle test had a better grade and recovery than the rougher cleaner with no recirculating
load. This may be due to differences in feed and assaying procedure.

Aluminum and red metals were then plotted individually to see how the circuits treated the
metals differently. Overall aluminum recovery was high with the highest aluminum recovery achieved at
97.69%. The test with the highest percentage of unrecoverable aluminum was Test 4, with about 4.5%
of the aluminum categorized as unrecoverable.

Red metals recovery was much lower. The highest red metals recovery achieved was 64.83% in
the RCS locked cycle test. In the tree analysis for the same test showed that only 69.8% of the red
metals in the feed were recoverable. Test 2 had the lowest amount of unrecoverable red metals, while
Test 4 had the largest percent of unrecoverable red metals at 42.12. The percent of metals that are
considered recoverable can be seen on all plots as the second leftmost point, if the range on all axis’s is
0 to 100.

The plot for red metals recovery vs. red metals grade was not easy to read, so red metals grade
was replaced with specific red metals grade. This calculation allowed the grade on the y-axis to
represent the relationship between of red metals mass and the mass of the waste. The grade vs.
recovery curves look conventional when specific grade is used, and still shows the same amount of
unrecoverable metal. It may be pertinent to use aluminum and red metal plots separately, instead of
zorba plots, if a processor prefers to create a product high in red metals.

When looking at all of the tests, most of the data seemed to be out of line with what would be
expected in circuit analysis. However, when looking at the plots, it was apparent that the test point
shifted within the grade recovery boundary in a manner that was consistent with circuit analysis. It was
realized that the test points needed to be compared to their respective tree analysis.

The separation efficiency of the circuits was compared to the point with the maximum
separation efficiency on the rejection/recovery boundary generated through tree analysis. The RCS
locked cycle test had the highest separation efficiency, and the second highest theoretical efficiency.
The rougher-scavenger in Test 1 had a lower separation efficiency, but a higher theoretical efficiency.
Comparing the theoretical efficiencies may be an indicator of which circuits performed closest to the
optimum point on the boundaries, but does not always fall in line with circuit analysis, and may not
make sense to use if there is a high target grade.

In another proposed way or comparing a test point to its corresponding rejection/recovery


boundary, recovery efficiency and rejection efficiency were compared between circuits. When these
values were compared, the results fell in line with what would be expected through circuit analysis.

67
However, as far as the author knows this is not a method that has been used in process engineering
before, and is cautious to make claims of how well it works without further testing.

Theoretical efficiency, recovery efficiency, and rejection efficiency were used in order to
minimize the differences in testing due to assays, and possible material changes. These tools would be
very valuable in the scrap recycling industry due to the fact that the feed to the shredder is constantly
changing, and there is no standard assaying procedure. Two different yards owned by the same
company could easily have different assaying procedures, and the quality of an assay can changed based
on the person who is conducting the assay. Further testing should be done on these values in order to
determine how useful they are.

Besides comparing circuits, tree analysis can also be used to see if machine settings need to be changed.
Changing machine settings should shift the boundary created by tree analysis. Just plotting a grade vs.
recovery curve is just scratching the surface of all the information that can be obtained through tree
analysis. Statistical programs can be used in order to see how different machine settings shift the tree
analysis boundary, if at all. Likewise could be done with changing size distributions instead of machine
settings. Like many other research projects, there are more ideas and questions for this topic at the end
than

68
REFERENCES
1. D.W. Brown, S.T. Hall, Froth flotation: preparation of a laboratory standard, Fuel, Volume 78,
Issue 14, November 1999, Pages 1621-1629, ISSN 0016-2361, http://dx.doi.org/10.1016/S0016-
2361(99)00108-8.

2. Pratten, S.J., Bensley, C.N. and Nicol, S.K., 1989. An evaluation of the flotation response of coals.
Int. J. Miner. Process., 27: 243-262

3. Shunli Zhang, Eric Forssberg, Bo Arvidson, William Moss, Aluminum recovery from electronic
scrap by High-Force® eddy-current separators, Resources, Conservation and Recycling, Volume
23, Issue 4, September 1998, Pages 225-241, ISSN 0921-3449, http://dx.doi.org/10.1016/S0921-
3449(98)00022-6.

4. T.P Meloy, D.A Whaley, M.C Williams, Flotation tree analysis — reexamined, International
Journal of Mineral Processing, Volume 55, Issue 1, October 1998, Pages 21-39, ISSN 0301-7516,
http://dx.doi.org/10.1016/S0301-7516(98)00023-4.

69
CHAPTER 4: CONCLUSIONS AND
RECCOMENDATIONS
Although density separation, commonly used in coal processing, has less variables than
separation of zorba from shredder residue some of the same analysis techniques can be used to
improve efficiency. In the Eriez RCS system it was clear that adding the cleaner and scavenger stages
with a recirculating load caused a major improvement in the separation of zorba from fluff. Using circuit
analysis, it is possible to meet a customers demand for high grade, without huge losses. Circuit analysis
is something that should be used for all separators in a scrap yard in order to make the best use of all of
the machines available.

Tree analysis is a good way to produce a grade vs. recovery curve in coal flotation, and it
appears that it can also be used to analyze circuits of eddy current separators. Using tree analysis, the
RCS locked cycle test had the best results in terms of combined rejection efficiency and recovery
efficiency, as well as overall separation efficiency. This backs up the results found from testing the Eriez
RCS system in that using a rougher, cleaner, and scavenger with a recirculating load, will give a better
performance than a single rougher.

The results from the tree analysis also back up some of the expectations of circuit analysis
theory stated in Chapter 2. The RCS test with no recirculation did not perform as well as the RCS Locked
cycle test. Importantly, this is still true when the tests compared to their respective tree analysis curves.
Circuit analysis states that there should be no difference in separation efficiency between a rougher only
and an RCS system with no recirculating load. Testing in Chapter 3 showed that there was a slight
improvement between using only the rougher and using the rougher-cleaner-scavenger with no
recirculation. However, the RCS locked cycle test, with a recirculating load, outperformed the RCS test
with no recirculating load in all measures including overall separation efficiency and theoretical
efficiency. The same is true for the rougher cleaner circuit when compared to the rougher cleaner
locked cycle circuit. This justifies adding recirculating streams in order to increase separation efficiency
if deemed economically feasible.

The recirculating load did not improve the separation in the rougher-scavenger circuit. When
looking at the tree analysis boundary for Test 4, it appears that the percentage of metals found in the
feed that were deemed unrecoverable is much larger than the other tests. This may indicate a change in
feed, a change in assaying procedure, or an error in conducting the test, among many other things.
There are reasons to believe that the feed for Test 4 contained a larger fraction of small, difficult to
recover, particles than previous tests, which may have made the rougher scavenger locked cycle test
perform worse than its counterpart in Test 1, even when compared to the tree analysis boundary.

Further testing has to be done on tree analysis, but if it is used in the field testing standards
have to be introduced. These standards should include, but are not limited to:

 A known and specific list of materials to be assayed for.


 A minimum assay size. Particles with a size less than a determined diameter should not
be counted in the assay at all.

70
 A branch in the tree analysis should stop when the mass of a sample is less than a
specified percentage of the mass of the feed.

Other standardizing methods that can be used for testing circuits or machine settings would be
the creation of a synthetic feed material that could have a known and exact contents and size
distribution.

The creation of standards was one of the goals of this research. Throughout the testing,
researchers involved gained a better understanding of sample size, and sampling technique. Standard
names and calculations have been introduced, yet not completely accepted, through the test work that
has been completed. Future standards, such as the standards for tree analysis, have been conceived yet
not completely realized.

After this testing has been completed, there are more questions about recycling and eddy
current separators than in the beginning of the test work. More testing should be done on tree analysis
and variations of tree analysis. It is especially important to see if changing machine settings can shift the
tree analysis boundary.

It is apparent through this research that a lack of a standard assay and small sample sizes can
skew recovery statistics reported. Other goals for this research also have not been met, but may be met
in the future.

71
APPENDIX A: TESTING OF ERIEZ RCS
SYSTEM

72
Test Plan:

RCS System:

Percent of feed recirculating from


Test Number Target Feed Rate (tph) Cleaner (%) Scavenger (%)
Test 1 3.25 10 30
Test 3 5 5 15
Test 4 1.5 5 30
Test 6 3.25 5 15
Test 8 3.25 5 15
Test 10 3.25 5 15
Test 12 5 10 15
Test 13 1.5 10 15
Test 16 5 5 30
Test 17 3.25 5 15
Test 18 3 20 20

Rougher Only:

Test Splitter Distance from Belt (in) Tumbleback Motor Frequency (Hz)
1 3.25 9
5 4 16
6 4 9
7 4.75 16
8 4.75 9
2 3.25 16
4 4 23

73
Assay Results:

RCS System

Zorba Contents Tailings Contents


Test Metal (g) Fluff (g) Metal (g) Fluff (g)
1 57608 1041 2434 631854.2
3 43091.3 376.8 9612 613166.166
4 34926.6 257.2 1378.921 334751
6 43045.92 442 4308 561547.4
8 39524 365 1414 503033.9
10 31987 349 932 368771
12 46773 994 3270 655441
13 37648.2 836 508 457221.1
16 61035.3893 1211.092 3762 685831.7
17 62798 693 4290 665420

Rougher Only

Zorba Contents Tailings Contents


Test Metal (g) Fluff (g) Metal (g) Fluff (g)
1 46878 30637 2371 422748
2 45041.72 39462.5 3889 496423.4
4 63625 23586.8 9531 816919.9
5 68123 27729 6948 782755.2811
6 80687 16356.541 8418 900127.5
7 65558 7711 16644 882829
8 72240 6231.5 14844 2126.27

74
Feed Rates Calculated:

RCS System

Feed Rate Total Feed


Test Time (min) (tph) Total Feed (lb) (tons) Volume (ft^3)
1 42 3.8 5320.000 2.660 161.212
3 42 3.9 5460.000 2.730 165.455
4 83.3 2.025 5622.750 2.811 170.386
6 42 3.33 4662.000 2.331 141.273
8 42 2.99 4186.000 2.093 126.848
10 42 2.2 3080.000 1.540 93.333
12 31.5 5.17 5428.500 2.714 164.500
13 84 1.36 3808.000 1.904 115.394
16 31.5 5.46 5733.000 2.867 173.727
17 42 4 5600.000 2.800 169.697
Total 48900.250 24.450 1481.826
Average 4890.025 2.445 148.183

Rougher Only

Feed Rate Total Feed


Test Time (min) (tph) Total Feed (lb) (tons) Volume (ft^3)
1 42 2.7 3780.000 1.890 114.545
2 31.5 4.3 4515.000 2.258 136.818
5 42 4.87 6818.000 3.409 206.606
6 84 2.77 7756.000 3.878 235.030
7 42 5.32 7448.000 3.724 225.697
8 84 2.9 8120.000 4.060 246.061
Total 38437.000 19.219 1164.758
Average 6406.167 3.203 194.126

75
Results:

RCS System

Test Feed Grade Grade of Yield to Zorba Grade Al Recovery Fluff


Number (%) Tailings (%) Zorba (%) (%) (%) Rejection (%)
Test 1 8.433 0.269 8.337 98.195 97.077 99.836
Test 3 7.941 1.581 6.522 99.098 81.388 99.936
Test 4 9.762 0.394 9.473 99.285 96.345 99.925
Test 6 7.744 0.787 7.120 98.499 90.555 99.884
Test 8 7.545 0.280 7.352 99.100 96.562 99.928
Test 10 8.076 0.249 7.934 98.897 97.157 99.905
Test 12 6.971 0.502 6.642 97.902 93.275 99.850
Test 13 7.694 0.110 7.763 97.799 98.682 99.815
Test 16 8.702 0.544 8.363 98.093 94.271 99.825
Test 17 9.243 0.628 8.767 98.889 93.797 99.893
Test 18 8.140 0.319 7.894 99.396 96.392 99.948

Rougher Only

Fluff
Feed Grade Grade of Yield To Zorba Grade Al Recovery Rejection
Test (%) Tailings (%) Zorba (%) (%) (%) (%)
1 9.819 0.549 15.464 60.491 95.273 93.225
2 7.874 0.777 14.449 49.899 91.563 92.142
3 9.165 0.776 22.360 38.296 93.429 84.811
4 8.073 1.148 9.638 72.998 87.154 97.169
5 8.367 0.869 10.677 71.097 90.720 96.632
6 8.911 0.901 9.746 83.093 90.877 98.191
7 7.778 1.834 6.780 89.499 78.015 99.228
8 8.299 1.480 7.534 91.996 83.510 99.342

76
RCS Test Results
100

90

80

70
Aluminum Recovery (%)

60

50

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Product Grade (%)

77
Rougher Only Test Results
100

90

80

70
Aluminum Recovery (%)

60

50

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Product Grade (%)

78
APPENDIX B: RAW DATA FOR TREE
ANALYSIS TESTS

79
Test 1:

Al Mass Brass Circuit Aluminum Wire


Sample Total (g) Cu (g) (g) ICW (g) Fluff (g) Boards (g) Only (g)
TTT,TTT 434.93 873.02 55.15 2328.64 213748.46 N/A N/A
Counted
TTT,TTC 22.6 10.21 0 in Cu 140.48 N/A N/A
Counted
TT,TTC 36.63 66.03 0 in Cu 171.58 N/A N/A
Counted
T,TTC 42.51 86.9 0 in Cu 203.09 N/A N/A
Counted
TTC 43.57 128.56 0 in Cu 211.92 N/A N/A
Counted
TCT 21.85 292.58 0 in Cu 181.14 N/A N/A
Counted
TCC 58.17 273.63 0 in Cu 75.71 N/A N/A
Counted
CTT 31.65 74.43 0 in Cu 1253.61 N/A N/A
Counted
CTC 345.46 65.22 0 in Cu 45.63 N/A N/A
Counted
CCT 29.63 104.2 0 in Cu 253.7 N/A N/A
Counted
C,CCT 330.02 50.8 0 in Cu 140.4 N/A N/A
Counted
CC,CCT 402.69 143.1 0 in Cu 96.84 N/A N/A
Counted
CCC,CCT 302.17 77.78 0 in Cu 42.52 N/A N/A
2441.5 2146.9 Counted
CCC,CCC 22998.02 1 2 in Cu 37.59 N/A N/A

Coning and quartering assay

Circuit Aluminum
Sample Al Mass (g) Cu (g) Brass (g) ICW (g) Fluff (g) Boards(g) Wire (g)
Not Not
TTT,TTT 59.23 118.89 7.51 317.12 29108.81301 Assayed Assayed

Sample Mass Assayed (g) Total Sample Mass (g) % Assayed


TTT,TTT 29676.52301 217917.20 13.62

80
Test 2:

Al Mass Brass Circuit Aluminum


Sample Total (g) Cu (g) (g) ICW (g) Fluff (g) Boards (g) Wire Only (g)
TTT,TTT 489.64 1570.11 0.00 3109.27 215730.29 N/A N/A
TTT,TTC 22.60 20.72 0.00 4.23 169.67 N/A N/A
TT,TTC 36.63 27.27 0.00 0.73 163.69 N/A N/A
T,TTC 42.51 45.36 2.67 2.97 178.53 N/A N/A
T,TCT 43.57 26.48 0.00 14.90 219.64 N/A N/A
T,TCC 21.85 21.22 0.00 2.21 8.50 N/A N/A
TCT 58.17 48.67 9.05 6.58 205.18 N/A N/A
TCC 55.10 46.97 8.03 7.31 16.54 N/A N/A
CTT 34.23 17.64 0.00 3.21 79.05 N/A N/A
C,TCT 20.20 7.11 10.10 0.00 6.68 N/A N/A
C,TCC 506.93 28.10 96.92 0.00 0.00 N/A N/A
C,CTT 31.65 8.50 0.00 18.53 29.46 N/A N/A
C,CTC 345.46 30.68 89.19 0.52 0.00 N/A N/A
CC,CTT 29.63 8.16 0.00 1.12 14.93 N/A N/A
CC,CTC 330.02 59.78 23.45 7.21 0.00 N/A N/A
CC,CCT 402.69 37.03 21.20 8.33 10.25 N/A N/A
CCC,CCT 302.17 28.81 8.98 0.00 13.40 N/A N/A
CCC,CCC 22138.46 1996.71 894.48 45.36 57.15 N/A N/A

Coning and quartering assay

Circuit Aluminum
Sample Al Mass (g) Cu (g) Brass (g) ICW (g) Fluff (g) Boards(g) Wire (g)
29206.788 Not Not
TTT,TTT 66.29 212.57 0 420.95 88 Assayed Assayed

Sample Mass Assayed (g) Total Sample Mass (g) % Assayed


TTT,TTT 29906.59888 220899.30 13.54

81
Test 3:

Al Mass Circuit Aluminum


Sample Total (g) Cu (g) Brass (g) ICW (g) Fluff (g) Boards (g) Wire Only (g)
TTT,TTT 832.61 1832.66 48.16 3146.43 221843.33 268.30 N/A
TTT,TTC 66.77 27.77 1.00 4.17 164.46 0.48 8.83
TT,TTC 89.86 34.37 0.00 8.26 170.23 2.41 0.00
T,TTC 85.75 52.53 3.16 11.35 168.89 2.29 0.00
T,TCT 75.19 37.59 8.02 10.88 186.51 2.18 11.34
T,TCC 63.30 27.77 1.00 4.17 9.88 1.18 8.90
T,CTT 53.40 30.04 3.57 8.51 191.91 0.95 7.09
T,CTC 41.56 20.71 7.14 0.70 6.59 0.33 6.89
T,CCT 51.33 21.59 5.10 14.20 25.00 0.75 5.28
T,CCC 250.26 80.65 21.32 7.54 16.54 1.31 18.94
C,TTT 14.75 3.63 2.53 10.51 66.71 0.00 1.11
C,TTC 15.73 2.15 23.42 0.00 0.19 0.50 0.93
C,TCT 38.05 8.55 7.31 2.04 15.28 0.00 1.09
C,TCC 410.63 31.13 72.80 1.13 0.03 2.60 4.90
C,CTT 26.60 7.73 2.75 0.07 53.96 0.67 0.00
C,CTC 526.18 40.18 108.43 0.00 0.30 7.34 4.28
C,CCT 366.36 30.15 22.43 5.60 16.58 1.57 9.45
CC,CCT 447.24 26.25 16.27 0.87 17.15 0.00 0.00
CCC,CCT 347.75 49.27 85.04 0.00 19.43 0.00 0.00
CCC,CCC 23072.41 2021.21 1249.19 31.75 59.88 59.87 0.00

Coning and quartering assay

Sample Al Mass (g) Cu (g) Brass (g) ICW (g) Fluff (g) Circuit Boards(g)
TTT,TTT 135.89 299.11 7.86 513.53 36207.16493 43.79

Sample Mass Assayed (g) Total Sample Mass (g) % Assayed


TTT,TTT 37163.55493 227703.18 16.32

82
Test 4:

Al Mass Brass Circuit Boards Aluminum


Sample Total (g) Cu (g) (g) ICW (g) Fluff (g) (g) Wire Only (g)
TTT,TTT 1196.87 2584.12 136.45 3913.79 223047.10 466.38 132.57
TTT,TTC 44.88 32.00 3.61 12.52 177.15 1.42 9.44
TT,TTC 46.96 41.69 0.12 1.72 183.76 1.32 8.17
T,TTC 60.24 47.23 2.33 12.24 208.64 0.73 13.01
T,TCT 54.64 43.68 2.24 7.35 178.95 2.28 10.07
T,TCC 26.74 17.40 0.00 0.00 9.08 0.60 7.84
T,CTT 48.80 41.15 2.65 7.42 217.59 1.31 6.70
T,CTC 18.35 14.45 2.80 1.68 5.69 2.15 5.40
T,CC 42.03 29.72 0.26 3.58 11.25 1.92 8.49
C,TTT 86.69 63.68 0.00 69.27 885.33 9.86 6.68
C,TTC 57.40 16.11 1.76 2.73 9.86 0.21 3.11
C,TC 240.69 46.47 4.89 2.56 11.82 1.76 7.65
C,CTT 37.64 26.16 1.04 11.38 103.04 2.20 2.12
C,CTC 207.92 55.20 11.80 3.21 1.01 3.28 5.28
C,CCT 498.56 41.93 20.20 12.39 34.81 7.50 7.50
CC,CCT 282.67 30.79 59.66 3.97 156.15 1.71 4.67
CCC,CCT 451.50 38.27 5.86 1.86 26.87 1.04 7.58
1515.0
CCC,CCC 22978.06 1517.72 0 31.75 32.66 59.87 70.76

Coning and quartering assay

Sample Al Mass (g) Cu (g) Brass (g) ICW (g) Fluff (g) Circuit Boards(g)
TTT,TTT 197.27 425.92 22.49 645.08 36763.11196 76.87

Sample Mass Assayed (g) Total Sample Mass (g) % Assayed


TTT,TTT 38053.87196 230878.33 16.48

83
APPENDIX C: TREE ANALYSIS DATA FOR
TEST 1

84
Tree Sample Diagram

85
Zorba Grade and Recovery:

Reds Mass Fluff Mass


Sample (g) (g) Zorba Metals (g) Total Mass (g)
TTT,TTT 3256.804642 213748.4608 3691.73551 217440.1963
TTT,TTC 10.21 140.48 32.81 173.29
TT,TTC 66.03 171.58 102.66 274.24
T,TTC 86.9 203.09 129.41 332.5
TTC 128.56 211.92 172.13 384.05
TCT 292.58 181.14 314.43 495.57
TCC 273.63 75.71 331.8 407.51
CTT 74.43 1253.61 106.08 1359.69
CTC 65.22 45.63 410.68 456.31
CCT 104.2 253.7 133.83 387.53
C,CCT 50.8 140.4 380.82 521.22
CC,CCT 143.1 96.84 545.79 642.63
CCC,CCT 77.78 42.52 379.95 422.47
CCC,CCC 4588.43 37.59 27586.45158 27624.04158

Zorba Total Mass Zorba Mass Zorba Yield Cumulative Grade Cumulative Zorba Mass
Sample Grade (%) (g) (g) (%) (%) (%)

CCC,CCC 99.8639 27624.0416 27586.4516 80.3834 99.8639 80.3834


CTC 90.0002 456.3100 410.6800 1.1967 99.7036 81.5801
CCC,CCT 89.9354 422.4700 379.9500 1.1071 99.5589 82.6872
CC,CCT 84.9307 642.6300 545.7900 1.5904 99.2363 84.2776
TCC 81.4213 407.5100 331.8000 0.9668 98.9907 85.2444
C,CCT 73.0632 521.2200 380.8200 1.1097 98.5413 86.3541
TCT 63.4482 495.5700 314.4300 0.9162 97.9724 87.2703
TTC 44.8197 384.0500 172.1300 0.5016 97.3129 87.7719
T,TTC 38.9203 332.5000 129.4100 0.3771 96.6924 88.1489
TT,TTC 37.4344 274.2400 102.6600 0.2991 96.1774 88.4481
CCT 34.5341 387.5300 133.8300 0.3900 95.4297 88.8380
TTT,TTC 18.9336 173.2900 32.8100 0.0956 95.0170 88.9336
CTT 7.8018 1359.6900 106.0800 0.3091 91.4751 89.2427
TTT,TTT 1.6978 217440.1963 3691.7355 10.7573 13.6770 100.0000

Circuit Grade (%) Recovery (%)


Rougher Only 94.046 86.086
Rougher Cleaner 98.071 84.581
Rougher Scavenger 93.418 87.969
Rougher Cleaner Scavenger 97.286 86.464

86
Test 1 Zorba Grade vs. Zorba Recovery
100

90 Tree
Analysis

80

Test 1
70 Rougher
Zorba Recovery (%)

60
Rougher
Cleaner
50 Scavenger

40 Rougher
Scavenger

30

Rougher
20 Cleaner

10

0
0 10 20 30 40 50 60 70 80 90 100
Zorba Grade (%)

87
Aluminum Grade and Recovery:

Sample Al Mass (g) Fluff Mass (g) Total Mass (g) Aluminum Grade (%)
TTT,TTT 434.931 213748.461 217440.196 0.200
TTT,TTC 22.600 140.480 173.290 13.042
TT,TTC 36.630 171.580 274.240 13.357
T,TTC 42.510 203.090 332.500 12.785
TTC 43.570 211.920 384.050 11.345
TCT 21.850 181.140 495.570 4.409
TCC 58.170 75.710 407.510 14.274
CTT 31.650 1253.610 1359.690 2.328
CTC 345.460 45.630 456.310 75.707
CCT 29.630 253.700 387.530 7.646
C,CCT 330.020 140.400 521.220 63.317
CC,CCT 402.690 96.840 642.630 62.663
CCC,CCT 302.170 42.520 422.470 71.525
CCC,CCC 22998.022 37.590 27624.042 83.254

Al
Grade Total Mass Cumulative Cumulative Al
Sample (%) (g) Al Mass (g) Al Yield (%) Grade (%) Mass (%)
CCC,CCC 83.254 27624.042 22998.022 91.626 83.254 91.626
CTC 75.707 456.310 345.460 1.376 83.131 93.002
CCC,CCT 71.525 422.470 302.170 1.204 82.959 94.206
C,CCT 63.317 521.220 330.020 1.315 82.606 95.521
CC,CCT 62.663 642.630 402.690 1.604 82.174 97.125
TCC 14.274 407.510 58.170 0.232 81.254 97.357
TT,TTC 13.357 274.240 36.630 0.146 80.641 97.503
TTT,TTC 13.042 173.290 22.600 0.090 80.257 97.593
T,TTC 12.785 332.500 42.510 0.169 79.530 97.762
TTC 11.345 384.050 43.570 0.174 78.691 97.936
CCT 7.646 387.530 29.630 0.118 77.821 98.054
TCT 4.409 495.570 21.850 0.087 76.688 98.141
CTT 2.328 1359.690 31.650 0.126 73.668 98.267
TTT,TTT 0.200 217440.196 434.931 1.733 10.003 100.000

Circuit Grade (%) Recovery (%)


Rougher Only 77.799 97.369
Rougher Cleaner 81.298 95.867
Rougher Scavenger 75.872 97.688
Rougher Cleaner Scavenger 79.153 96.186

88
Test 1 Aluminum Grade vs. Aluminum Recovery
100

90

Tree
80 Analysis

70 Test 1
Rougher
Aluminum Recovery (%)

60
Rougher
Cleaner
Scavenger
50
Rougher
Scavenger
40

Rougher
30 Cleaner

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Aluminum Grade (%)

89
Red Metals Grade and Recovery:

Reds Mass
Sample (g) Fluff Mass (g) Total Mass (g) Reds Grade (%)
TTT,TTT 3256.805 213748.461 217440.196 1.498
TTT,TTC 10.210 140.480 173.290 5.892
TT,TTC 66.030 171.580 274.240 24.077
T,TTC 86.900 203.090 332.500 26.135
TTC 128.560 211.920 384.050 33.475
TCT 292.580 181.140 495.570 59.039
TCC 273.630 75.710 407.510 67.147
CTT 74.430 1253.610 1359.690 5.474
CTC 65.220 45.630 456.310 14.293
CCT 104.200 253.700 387.530 26.888
C,CCT 50.800 140.400 521.220 9.746
CC,CCT 143.100 96.840 642.630 22.268
CCC,CCT 77.780 42.520 422.470 18.411
CCC,CCC 4588.430 37.590 27624.042 16.610

Cumulative Cumulative
Sample Reds Grade (%) Total Mass (g) Reds Mass (g) Reds Yield (%) Grade (%) Reds Mass (%)

TCC 67.147 407.510 273.630 2.968 67.147 2.968


TCT 59.039 495.570 292.580 3.174 62.698 6.142
TTC 33.475 384.050 128.560 1.395 53.978 7.537
CCT 26.888 387.530 104.200 1.130 47.709 8.667
T,TTC 26.135 332.500 86.900 0.943 44.135 9.610
TT,TTC 24.077 274.240 66.030 0.716 41.724 10.326
CC,CCT 22.268 642.630 143.100 1.552 37.448 11.878
CCC,CCT 18.411 422.470 77.780 0.844 35.045 12.722
CCC,CCC 16.610 27624.042 4588.430 49.773 18.602 62.495
CTC 14.293 456.310 65.220 0.707 18.540 63.202
C,CCT 9.746 521.220 50.800 0.551 18.396 63.754
TTT,TTC 5.892 173.290 10.210 0.111 18.329 63.864
CTT 5.474 1359.690 74.430 0.807 17.807 64.672
TTT,TTT 1.498 217440.196 3256.805 35.328 3.674 100.000

Circuit Grade (%) Recovery (%)


Rougher Only 16.247 55.365
Rougher Cleaner 16.773 53.851
Rougher Scavenger 17.545 61.507
Rougher Cleaner Scavenger 18.132 59.993

90
Test 1 Reds Grade vs. Reds Recovery
100

90 Tree Analysis

80

70 Test 1 Rougher

60
Recovery (%)

Rougher
50 Cleaner
Scavenger

40
Rougher
30 Scavenger

20
Rougher
Cleaner
10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

91
Specific Reds Grade vs. Recovery

Sample Reds Mass (g) Fluff Mass (g) Total Relevant Mass (g) Specific Reds Grade (%)
TTT,TTT 3256.805 213748.461 217005.265 1.501
TTT,TTC 10.210 140.480 150.690 6.775
TT,TTC 66.030 171.580 237.610 27.789
T,TTC 86.900 203.090 289.990 29.967
TTC 128.560 211.920 340.480 37.758
TCT 292.580 181.140 473.720 61.762
TCC 273.630 75.710 349.340 78.328
CTT 74.430 1253.610 1328.040 5.604
CTC 65.220 45.630 110.850 58.836
CCT 104.200 253.700 357.900 29.114
C,CCT 50.800 140.400 191.200 26.569
CC,CCT 143.100 96.840 239.940 59.640
CCC,CCT 77.780 42.520 120.300 64.655
CCC,CCC 4588.430 37.590 4626.020 99.187

Cumulative
Specific Reds Total Relevant Specific Grade Cumulative Reds
Sample Grade (%) Mass (g) Reds Mass (g) Reds Yield (%) (%) Mass (%)

CCC,CCC 99.187 4626.020 4588.430 49.773 99.187 49.773


TCC 78.328 349.340 273.630 2.968 97.723 52.741
CCC,CCT 64.655 120.300 77.780 0.844 96.942 53.585
TCT 61.762 473.720 292.580 3.174 93.950 56.759
CC,CCT 59.640 239.940 143.100 1.552 92.533 58.311
CTC 58.836 110.850 65.220 0.707 91.902 59.019
TTC 37.758 340.480 128.560 1.395 88.957 60.413
T,TTC 29.967 289.990 86.900 0.943 86.346 61.356
CCT 29.114 357.900 104.200 1.130 83.381 62.486
TT,TTC 27.789 237.610 66.030 0.716 81.532 63.202
C,CCT 26.569 191.200 50.800 0.551 80.100 63.754
TTT,TTC 6.775 150.690 10.210 0.111 78.625 63.864
CTT 5.604 1328.040 74.430 0.807 67.625 64.672
TTT,TTT 1.501 217005.265 3256.805 35.328 4.082 100.000

Circuit Specific Grade (%) Recovery (%)


Rougher Only 73.183 55.365
Rougher Cleaner 89.684 53.851
Rougher Scavenger 72.720 61.507
Rougher Cleaner
Scavenger 86.979 59.993

92
Specific Reds Grade vs. Reds Recovery
100

90
Tree
Analysis
80

70 Rougher

60
Recovery (%)

Rougher
50 Cleaner
Scavenger

40
Rougher
Scavenger
30

Rougher
20 Cleaner

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

93
Zorba Recovery and Fluff Rejection:

Sample Zorba Mass (g) Fluff Mass (g) Total Mass (g) Zorba Grade (%)
TTT,TTT 3691.736 213748.461 217440.196 1.698
TTT,TTC 32.810 140.480 173.290 18.934
TT,TTC 102.660 171.580 274.240 37.434
T,TTC 129.410 203.090 332.500 38.920
TTC 172.130 211.920 384.050 44.820
TCT 314.430 181.140 495.570 63.448
TCC 331.800 75.710 407.510 81.421
CTT 106.080 1253.610 1359.690 7.802
CTC 410.680 45.630 456.310 90.000
CCT 133.830 253.700 387.530 34.534
C,CCT 380.820 140.400 521.220 73.063
CC,CCT 545.790 96.840 642.630 84.931
CCC,CCT 379.950 42.520 422.470 89.935
CCC,CCC 27586.452 37.590 27624.042 99.864

Zorba Yield Cumulative Fluff Cumulative Zorba Separation


Sample Zorba Grade (%) Fluff Yield (%) (%) Yield to T (%) Mass (%) Efficiency (%)

CCC,CCC 99.864 0.017 80.383 99.983 80.383 80.366


CTC 90.000 0.021 1.197 99.962 81.580 81.542
CCC,CCT 89.935 0.020 1.107 99.942 82.687 82.629
CC,CCT 84.931 0.045 1.590 99.897 84.278 84.175
TCC 81.421 0.035 0.967 99.862 85.244 85.107
C,CCT 73.063 0.065 1.110 99.797 86.354 86.152
TCT 63.448 0.084 0.916 99.714 87.270 86.984
TTC 44.820 0.098 0.502 99.616 87.772 87.388
T,TTC 38.920 0.094 0.377 99.522 88.149 87.671
TT,TTC 37.434 0.079 0.299 99.443 88.448 87.891
CCT 34.534 0.117 0.390 99.326 88.838 88.164
TTT,TTC 18.934 0.065 0.096 99.261 88.934 88.195
CTT 7.802 0.579 0.309 98.682 89.243 87.925
TTT,TTT 1.698 98.682 10.757 0.000 100.000 0.000

Fluff Rejection Zorba Recovery Separation Theoretical


Circuit (%) (%) Efficiency (%) Efficiency (%)
Rougher Only 99.137 86.086 85.223 96.927
Rougher Cleaner 99.736 84.581 84.317 95.896
Rougher Scavenger 99.018 87.969 86.987 98.933
Rougher Cleaner
Scavenger 99.618 86.464 86.081 97.903

94
Test 1 R: Fluff Rejection vs. Zorba Recovery
100

90

80

70
Zorba Recovery (%)

60
Tree Analysis
Test 1 Rougher
50
SE = 80
SE = 90
40
SE = 88.2
SE = 85.2
30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Fluff Rejection (%)

95
Test 1 R-C: Fluff Rejection vs. Zorba Recovery
100

90
Tree Analysis

80
Test 1
Rougher
70
Cleaner

SE = 80
Zorba Recovery (%)

60

50 SE = 90

40
SE = 88.2

30
SE = 84.3
20

10

0
0 10 20 30 40 50 60 70 80 90 100
Fluff Rejection (%)

96
Test 1 R-S: Fluff Rejection vs. Zorba Recovery
100

90
Tree
Analysis
80

Test 1
Rougher
70
Scavenger

SE = 80
Zorba Recovery (%)

60

50 SE = 90

40
SE = 87

30

SE= 88.2
20

10

0
0 10 20 30 40 50 60 70 80 90 100
Fluff Rejection (%)

97
Test 1 R-C-S: Fluff Rejection vs. Zorba Recovery
100

90 Tree Analysis

80
Test 1 Rougher
Cleaner
Scavenger
70
SE = 80
Zorba Recovery (%)

60
SE = 90
50

SE = 88.2
40

SE = 86
30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Fluff Rejection (%)

98
Rejection Efficiency, Recovery Efficiency, and Spline Fitting

Fluff Cumulative
Sample Rejection (x) Zorba Mass (y) s dy dx m
CCC,CCC 99.983 80.383 -56.8051 1.1967 -0.0211
CTC 99.962 81.580 -56.3985 1.1071 -0.0196 -56.6018
CCC,CCT 99.942 82.687 -35.5718 1.5904 -0.0447 -45.9851
CC,CCT 99.897 84.278 -27.6603 0.9668 -0.0350 -31.6161
TCC 99.862 85.244 -17.1193 1.1097 -0.0648 -22.3898
C,CCT 99.797 86.354 -10.9558 0.9162 -0.0836 -14.0376
TCT 99.714 87.270 -5.1265 0.5016 -0.0978 -8.0411
TTC 99.616 87.772 -4.0217 0.3771 -0.0938 -4.5741
T,TTC 99.522 88.149 -3.7763 0.2991 -0.0792 -3.8990
TT,TTC 99.443 88.448 -3.3294 0.3900 -0.1171 -3.5529
CCT 99.326 88.838 -1.4741 0.0956 -0.0649 -2.4018
TTT,TTC 99.261 88.934 -0.5341 0.3091 -0.5788 -1.0041
CTT 98.682 89.243 -0.1090 10.7573 -98.6723 -0.3215
TTT,TTT 0.010 100.000 0.0000 0.0000 -0.0090 -0.0545

Sample a b c d
CCC,CCC 80.3834 0.0000 5402.6575 128459.3964
CTC 81.5801 -56.6018 509.7583 26495.3719
CCC,CCT 82.6872 -45.9851 -377.3561 -3230.6839
CC,CCT 84.2776 -31.6161 -75.5551 1076.1681
TCC 85.2444 -22.3898 -115.0776 -520.9373
C,CCT 86.3541 -14.0376 -38.8493 -23.8949
TCT 87.2703 -8.0411 -53.9355 -246.7841
TTC 87.7719 -4.5741 -10.4737 -48.8740
T,TTC 88.1489 -3.8990 -0.2772 16.0557
TT,TTC 88.4481 -3.5529 4.1045 51.3315
CCT 88.8380 -2.4018 -21.3597 -108.8003
TTT,TTC 88.9336 -1.0041 -1.2570 -0.7687
CTT 89.2427 -0.3215 -0.0038 0.0000
TTT,TTT 100.0000 -0.0545 5555543.4422 617283277.6544

99
Fluff Zorba Efficient Efficient
Rejection Recovery Rejection Recovery Rejection Recovery
Test (%) (%) (%) (%) Efficiency (%) Efficiency
Rougher 99.137 86.086 99.816 89.068 99.319 96.652
Rougher
Cleaner 99.736 84.581 99.888 87.066 99.849 95.577
Rougher
Scavenger 99.018 87.969 99.569 89.124 99.447 98.704
Rougher
Cleaner
Scavenger 99.618 86.464 99.784 87.764 99.834 97.523

100
Test 1: Fluff Rejection vs. Zorba Recovery
100

90

80
Tree
Analysis
70

Test 1
60 Rougher
Recovery (%)

50 Efficent
Rejection

40
Efficent
Recovery
30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

101
Test 1: Fluff Rejection vs. Zorba Recovery
100

90

80
Tree
Analysis
70
Rougher
Cleaner
60
Recovery (%)

Efficent
50 Rejection

Efficent
40 Recovery

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

102
Test 1: Fluff Rejection vs. Zorba Recovery
100

90

Tree
80 Analysis

70 Rougher
Scavenger

60
Recovery (%)

Efficent
Rejection
50

Efficent
40 Recovery

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

103
Test 1: Fluff Rejection vs. Zorba Recovery
100

90

80
Tree
Analysis
70
Rougher
Cleaner
60 Scavenger
Recovery (%)

Efficent
Rejection
50

Efficent
40 Recovery

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

104
APPENDIX D: TREE ANALYSIS FOR TEST 2

105
Tree Sample Diagram:

106
Zorba Grade and Recovery:

Sample Reds Mass (g) Fluff Mass (g) Zorba Metals (g) Total Mass (g)
TTT,TTT 1570.11 218839.56 2059.75 220899.30
TTT,TTC 20.72 173.90 43.32 217.22
TT,TTC 27.27 164.42 63.90 228.32
T,TTC 48.03 181.50 90.54 272.04
T,TCT 26.48 234.54 70.05 304.59
T,TCC 21.22 10.71 43.07 53.78
TCT 57.72 211.76 115.89 327.65
TCC 55.00 23.85 110.10 133.95
CTT 17.64 82.26 51.87 134.13
C,TCT 17.21 6.68 37.41 44.09
C,TCC 125.02 0.00 631.95 631.95
C,CTT 8.50 47.99 40.15 88.14
C,CTC 119.87 0.52 465.33 465.85
CC,CTT 8.16 16.05 37.79 53.84
CC,CTC 83.23 7.21 413.25 420.46
CC,CCT 58.23 18.58 460.92 479.50
CCC,CCT 37.79 13.40 339.96 353.36
CCC,CCC 2891.20 102.51 25029.66 25132.17

107
Cumulative
Zorba Grade Total Mass Zorba Mass Zorba Yield Cumulative Zorba Mass
Sample (%) (g) (g) (%) Grade (%) (%)
C,TCC 100.000 631.950 631.950 2.099 100.000 2.099
C,CTC 99.888 465.850 465.330 1.546 99.953 3.645
CCC,CCC 99.592 25132.172 25029.660 83.141 99.607 86.786
CC,CTC 98.285 420.460 413.250 1.373 99.586 88.159
CCC,CCT 96.208 353.360 339.960 1.129 99.542 89.288
CC,CCT 96.125 479.500 460.920 1.531 99.483 90.819
C,TCT 84.849 44.090 37.410 0.124 99.459 90.944
TCC 82.195 133.950 110.100 0.366 99.375 91.309
T,TCC 80.086 53.780 43.070 0.143 99.338 91.452
CC,CTT 70.189 53.840 37.790 0.126 99.282 91.578
C,CTT 45.553 88.140 40.150 0.133 99.112 91.711
CTT 38.671 134.130 51.870 0.172 98.822 91.884
TCT 35.370 327.650 115.890 0.385 98.088 92.269
T,TTC 33.282 272.040 90.540 0.301 97.471 92.569
TT,TTC 27.987 228.320 63.900 0.212 96.921 92.782
T,TCT 22.998 304.590 70.050 0.233 96.148 93.014
TTT,TTC 19.943 217.220 43.320 0.144 95.583 93.158
TTT,TTT 0.932 220899.304 2059.745 6.842 12.030 100.000

Recovery
Circuit Grade (%) (%)
RCS Locked Cycle 98.938 91.375

108
Test 2 Zorba Grade vs. Zorba Recovery
100
95
90
85
80
75
70
65 Tree Analysis

60
Recovery (%)

55
50
45
Test 2 RCS
40
Locked Cycle
35
30
25
20
15
10
5
0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

109
Aluminum Grade and Recovery:

Sample Al Mass (g) Fluff Mass (g) Total Mass (g) Aluminum Grade (%)
TTT,TTT 489.638 218839.559 220899.304 0.222
TTT,TTC 22.600 173.900 217.220 10.404
TT,TTC 36.630 164.420 228.320 16.043
T,TTC 42.510 181.500 272.040 15.626
T,TCT 43.570 234.540 304.590 14.304
T,TCC 21.850 10.710 53.780 40.628
TCT 58.170 211.760 327.650 17.754
TCC 55.100 23.850 133.950 41.135
CTT 34.230 82.260 134.130 25.520
C,TCT 20.200 6.680 44.090 45.815
C,TCC 506.930 0.000 631.950 80.217
C,CTT 31.650 47.990 88.140 35.909
C,CTC 345.460 0.520 465.850 74.157
CC,CTT 29.630 16.050 53.840 55.033
CC,CTC 330.020 7.210 420.460 78.490
CC,CCT 402.690 18.580 479.500 83.981
CCC,CCT 302.170 13.400 353.360 85.513
CCC,CCC 22138.465 102.512 25132.172 88.088

110
Total Mass Cumulative Cumulative
Sample Al Grade (%) (g) Al Mass (g) Al Yield (%) Grade (%) Al Mass (%)
CCC,CCC 88.088 25132.172 22138.465 88.868 88.088 88.868
CCC,CCT 85.513 353.360 302.170 1.213 88.052 90.081
CC,CCT 83.981 479.500 402.690 1.616 87.977 91.698
C,TCC 80.217 631.950 506.930 2.035 87.793 93.733
CC,CTC 78.490 420.460 330.020 1.325 87.648 95.058
C,CTC 74.157 465.850 345.460 1.387 87.419 96.444
CC,CTT 55.033 53.840 29.630 0.119 87.356 96.563
C,TCT 45.815 44.090 20.200 0.081 87.290 96.644
TCC 41.135 133.950 55.100 0.221 87.067 96.866
T,TCC 40.628 53.780 21.850 0.088 86.977 96.953
C,CTT 35.909 88.140 31.650 0.127 86.815 97.080
CTT 25.520 134.130 34.230 0.137 86.521 97.218
TCT 17.754 327.650 58.170 0.234 85.726 97.451
TT,TTC 16.043 228.320 36.630 0.147 85.168 97.598
T,TTC 15.626 272.040 42.510 0.171 84.512 97.769
T,TCT 14.304 304.590 43.570 0.175 83.778 97.944
TTT,TTC 10.404 217.220 22.600 0.091 83.235 98.034
TTT,TTT 0.222 220899.304 489.638 1.966 9.955 100.000

Circuit Al Grade (%) Recovery (%)


RCS Locked Cycle 86.829 96.909

111
Test 2 Aluminum Grade vs. Aluminum Recovery
100

90

80

70 Tree
Analysis

60
Recovery (%)

50 Test 2 RCS
Locked Cycle

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

112
Red Metals Grade and Recovery:

Sample Reds Mass (g) Fluff Mass (g) Total Mass (g) Reds Grade (%)
TTT,TTT 1570.107 218839.559 220899.304 0.711
TTT,TTC 20.720 173.900 217.220 9.539
TT,TTC 27.270 164.420 228.320 11.944
T,TTC 48.030 181.500 272.040 17.655
T,TCT 26.480 234.540 304.590 8.694
T,TCC 21.220 10.710 53.780 39.457
TCT 57.720 211.760 327.650 17.616
TCC 55.000 23.850 133.950 41.060
CTT 17.640 82.260 134.130 13.151
C,TCT 17.210 6.680 44.090 39.034
C,TCC 125.020 0.000 631.950 19.783
C,CTT 8.500 47.990 88.140 9.644
C,CTC 119.870 0.520 465.850 25.731
CC,CTT 8.160 16.050 53.840 15.156
CC,CTC 83.230 7.210 420.460 19.795
CC,CCT 58.230 18.580 479.500 12.144
CCC,CCT 37.790 13.400 353.360 10.694
CCC,CCC 2891.195 102.512 25132.172 11.504

113
Cumulative
Reds Grade Reds Mass Reds Yield Cumulative Reds Mass
Sample (%) Total Mass (g) (g) (%) Grade (%) (%)
TCC 41.060 133.950 55.000 1.059 41.060 1.059
T,TCC 39.457 53.780 21.220 0.409 40.601 1.468
C,TCT 39.034 44.090 17.210 0.331 40.303 1.799
C,CTC 25.731 465.850 119.870 2.308 30.573 4.107
CC,CTC 19.795 420.460 83.230 1.603 26.520 5.710
C,TCC 19.783 631.950 125.020 2.407 24.087 8.117
T,TTC 17.655 272.040 48.030 0.925 23.222 9.042
TCT 17.616 327.650 57.720 1.111 22.440 10.153
CC,CTT 15.156 53.840 8.160 0.157 22.277 10.310
CTT 13.151 134.130 17.640 0.340 21.795 10.650
CC,CCT 12.144 479.500 58.230 1.121 20.261 11.771
TT,TTC 11.944 228.320 27.270 0.525 19.676 12.296
CCC,CCC 11.504 25132.172 2891.195 55.671 12.439 67.967
CCC,CCT 10.694 353.360 37.790 0.728 12.417 68.695
C,CTT 9.644 88.140 8.500 0.164 12.409 68.858
TTT,TTC 9.539 217.220 20.720 0.399 12.387 69.257
T,TCT 8.694 304.590 26.480 0.510 12.349 69.767
TTT,TTT 0.711 220899.304 1570.107 30.233 2.075 100.000

Circuit Grade (%) Recovery (%)


RCS Locked Cycle 12.109 64.829

114
Test 2 Reds Grade vs. Reds Recovery
100

90

80
Tree
Analysis
70

60
Recovery (%)

Test 2 RCS
50 Locked
Cycle

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

115
Specific Reds Grade and Recovery:

Total Relevant Specific Reds


Sample Reds Mass (g) Fluff Mass (g) Mass (g) Grade (%)
TTT,TTT 1570.107 218839.559 220409.666 0.712
TTT,TTC 20.720 173.900 194.620 10.646
TT,TTC 27.270 164.420 191.690 14.226
T,TTC 48.030 181.500 229.530 20.925
T,TCT 26.480 234.540 261.020 10.145
T,TCC 21.220 10.710 31.930 66.458
TCT 57.720 211.760 269.480 21.419
TCC 55.000 23.850 78.850 69.753
CTT 17.640 82.260 99.900 17.658
CTCT 17.210 6.680 23.890 72.039
CTCC 125.020 0.000 125.020 100.000
CCTT 8.500 47.990 56.490 15.047
CCTC 119.870 0.520 120.390 99.568
CC,CTT 8.160 16.050 24.210 33.705
CC,CTC 83.230 7.210 90.440 92.028
CC,CCT 58.230 18.580 76.810 75.810
CCC,CCT 37.790 13.400 51.190 73.823
CCC,CCC 2891.195 102.512 2993.707 96.576

116
Specific Total Cumulative Cumulative
Reds Grade Relevant Reds Mass Reds Yield Specific Reds Mass
Sample (%) Mass (g) (g) (%) Grade (%) (%)
CTCC 100.000 125.020 125.020 2.407 100.000 2.099
CCTC 99.568 120.390 119.870 2.308 99.788 3.645
CCC,CCC 96.576 2993.707 2891.195 55.671 96.819 86.786
CC,CTC 92.028 90.440 83.230 1.603 96.689 88.159
CC,CCT 75.810 76.810 58.230 1.121 96.218 89.288
CCC,CCT 73.823 51.190 37.790 0.728 95.887 90.819
CTCT 72.039 23.890 17.210 0.331 95.723 90.944
TCC 69.753 78.850 55.000 1.059 95.148 91.309
T,TCC 66.458 31.930 21.220 0.409 94.893 91.452
CC,CTT 33.705 24.210 8.160 0.157 94.483 91.578
TCT 21.419 269.480 57.720 1.111 89.416 91.711
T,TTC 20.925 229.530 48.030 0.925 85.596 91.884
CTT 17.658 99.900 17.640 0.340 83.986 92.269
CCTT 15.047 56.490 8.500 0.164 83.075 92.569
TT,TTC 14.226 191.690 27.270 0.525 80.118 92.782
TTT,TTC 10.646 194.620 20.720 0.399 77.215 93.014
T,TCT 10.145 261.020 26.480 0.510 73.656 93.158
TTT,TTT 0.712 220409.666 1570.107 30.233 2.305 100.000

Circuit Specific Grade (%) Recovery (%)


RCS Locked Cycle 91.939 64.829

117
Specific Reds Grade vs. Specific Reds Recovery
100

90

80

Tree Analysis
70

60
Recovery (%)

50
Rougher
Cleaner
Scavenger
40
Locked Cycle

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

118
Zorba Recovery and Fluff Rejection:

Sample Zorba Mass (g) Fluff Mass (g) Total Mass (g) Zorba Grade (%)
TTT,TTT 2059.745 218839.559 220899.304 0.932
TTT,TTC 43.320 173.900 217.220 19.943
TT,TTC 63.900 164.420 228.320 27.987
T,TTC 90.540 181.500 272.040 33.282
T,TCT 70.050 234.540 304.590 22.998
T,TCC 43.070 10.710 53.780 80.086
TCT 115.890 211.760 327.650 35.370
TCC 110.100 23.850 133.950 82.195
CTT 51.870 82.260 134.130 38.671
CTCT 37.410 6.680 44.090 84.849
CTCC 631.950 0.000 631.950 100.000
CCTT 40.150 47.990 88.140 45.553
CCTC 465.330 0.520 465.850 99.888
CC,CTT 37.790 16.050 53.840 70.189
CC,CTC 413.250 7.210 420.460 98.285
CC,CCT 460.920 18.580 479.500 96.125
CCC,CCT 339.960 13.400 353.360 96.208
CCC,CCC 25029.660 102.512 25132.172 99.592

119
Cumulative Cumulative Separation
Zorba Grade Fluff Yield Zorba Yield Fluff Yield to Zorba Mass Efficiency
Sample (%) (%) (%) T (%) (%) (%)
CTCC 100.000 0.000 2.099 100.000 2.099 2.099
CCTC 99.888 0.000 1.546 100.000 3.645 3.645
CCC,CCC 99.592 0.047 83.141 99.953 86.786 86.740
CC,CTC 98.285 0.003 1.373 99.950 88.159 88.109
CCC,CCT 96.208 0.006 1.129 99.944 89.288 89.232
CC,CCT 96.125 0.008 1.531 99.935 90.819 90.755
CTCT 84.849 0.003 0.124 99.932 90.944 90.876
TCC 82.195 0.011 0.366 99.922 91.309 91.231
T,TCC 80.086 0.005 0.143 99.917 91.452 91.369
CC,CTT 70.189 0.007 0.126 99.909 91.578 91.487
CCTT 45.553 0.022 0.133 99.888 91.711 91.599
CTT 38.671 0.037 0.172 99.850 91.884 91.734
TCT 35.370 0.096 0.385 99.754 92.269 92.023
T,TTC 33.282 0.082 0.301 99.672 92.569 92.241
TT,TTC 27.987 0.075 0.212 99.597 92.782 92.378
T,TCT 22.998 0.107 0.233 99.490 93.014 92.505
TTT,TTC 19.943 0.079 0.144 99.411 93.158 92.569
TTT,TTT 0.932 99.411 6.842 0.000 100.000 0.000

Fluff Rejection Zorba Recovery Separation Efficiency Theoretical Efficiency


Circuit (%) (%) (%) (%)
RCS Locked
Cycle 99.866 91.375 91.241 98.565

120
Test 2: Fluff Rejection vs. Zorba Recovery
100

90

Tree
80 Analysis

SE = 80
70

60 SE = 90
Recovery (%)

50 SE = 92.5

40 SE = 91.25

30
RCS
Locked
20 Cycle

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

121
Rejection Efficiency, Recovery Efficiency, and Spline Fitting

Cumulative
Fluff Zorba Mass
Sample Rejection (x) (y) s dy dx m
C,TCC 100.000 2.099 -6543.5045 1.5457 -0.0002
C,CTC 100.000 3.645 -1785.3930 83.1415 -0.0466 -4164.4488
CCC,CCC 99.953 86.786 -419.1122 1.3727 -0.0033 -1102.2526
CC,CTC 99.950 88.159 -185.5136 1.1293 -0.0061 -302.3129
CCC,CCT 99.944 89.288 -181.3980 1.5310 -0.0084 -183.4558
CC,CCT 99.935 90.819 -40.9509 0.1243 -0.0030 -111.1745
C,TCT 99.932 90.944 -33.7561 0.3657 -0.0108 -37.3535
TCC 99.922 91.309 -29.4061 0.1431 -0.0049 -31.5811
T,TCC 99.917 91.452 -17.2169 0.1255 -0.0073 -23.3115
CC,CTT 99.909 91.578 -6.1177 0.1334 -0.0218 -11.6673
C,CTT 99.888 91.711 -4.6108 0.1723 -0.0374 -5.3643
CTT 99.850 91.884 -4.0018 0.3850 -0.0962 -4.3063
TCT 99.754 92.269 -3.6477 0.3007 -0.0824 -3.8247
T,TTC 99.672 92.569 -2.8418 0.2123 -0.0747 -3.2448
TT,TTC 99.597 92.782 -2.1840 0.2327 -0.1065 -2.5129
T,TCT 99.490 93.014 -1.8216 0.1439 -0.0790 -2.0028
TTT,TTC 99.411 93.158 -0.0688 6.8419 -99.4113 -0.9452
TTT,TTT 0.000 100.000

122
Sample a b c d
C,TCC 2.0992 0.0000 65473634.3160 159905255229.3940
C,CTC 3.6449 -4164.4488 -87506.5733 -782053.4885
CCC,CCC 86.7863 -1102.2526 -381491.2049 -52794354.0003
CC,CTC 88.1590 -302.3129 -38037.5680 -3096641.5480
CCC,CCT 89.2883 -183.4558 7832.4581 956873.9985
CC,CCT 90.8193 -111.1745 -45097.9828 -7235546.3134
C,TCT 90.9436 -37.3535 -463.3406 -12118.6404
TCC 91.3093 -31.5811 358.6142 165596.5808
T,TCC 91.4524 -23.3115 -910.6701 -10253.0568
CC,CTT 91.5779 -11.6673 -474.5719 -10091.9068
C,CTT 91.7113 -5.3643 -32.1753 -321.4788
CTT 91.8836 -4.3063 -4.4908 -13.7749
TCT 92.2685 -3.8247 0.5920 33.2256
T,TTC 92.5693 -3.2448 -6.3851 -13.2616
TT,TTC 92.7815 -2.5129 -4.4740 -13.0148
T,TCT 93.0142 -2.0028 6.5061 111.3953
TTT,TTC 93.1581 -0.9452
TTT,TTT

Fluff Zorba Efficient Efficient Rejection


Rejection Recovery Rejection Recovery Efficiency Recovery
Test (%) (%) (%) (%) (%) Efficiency
RCS Locked
Cycle 99.866 91.375 99.920 91.815 99.946 99.520

123
Test 2: Fluff Rejection vs. Zorba Recovery
100

90

80
Tree
Analysis
70
RCS
Locked
60 Cycle
Recovery (%)

Efficent
Rejection
50

Efficent
40 Recovery

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

124
APPENDIX E: TREE ANALYSIS DATA FOR
TEST 3

125
Tree Sample Diagram:

126
Zorba Grade and Recovery:

Sample Reds Mass (g) Fluff Mass (g) Zorba Metals (g) Total Mass (g)
TTT,TTT 1880.82 225258.06 2713.43 227971.49
TTT,TTC 28.77 169.11 95.54 264.65
TT,TTC 34.37 180.90 124.23 305.13
T,TTC 55.69 182.53 141.44 323.96
T,TCT 45.61 199.57 120.80 320.37
T,TCC 28.77 15.23 92.07 107.30
T,CTT 33.61 201.37 87.01 288.38
T,CTC 27.85 7.62 69.41 77.03
T,CCT 26.69 39.95 78.02 117.97
T,CCC 101.97 25.39 352.23 377.62
C,TTT 6.16 77.22 20.91 98.13
C,TTC 25.57 0.69 41.30 41.99
C,TCT 15.86 17.32 53.91 71.23
C,TCC 103.93 3.76 514.56 518.32
C,CTT 10.48 54.70 37.08 91.78
C,CTC 148.61 7.64 674.79 682.43
C,CCT 52.58 23.75 418.94 442.69
CC,CCT 42.52 18.02 489.76 507.78
CCC,CCT 134.31 19.43 482.06 501.49
CCC,CCC 3270.40 151.51 26342.81 26494.31

127
Cumulative
Zorba Grade Total Mass Zorba Mass Zorba Yield Cumulative Zorba Mass
Sample (%) (g) (g) (%) Grade (%) (%)
CCC,CCC 99.428 26494.315 26342.809 79.947 99.428 79.947
C,TCC 99.275 518.320 514.560 1.562 99.425 81.509
C,CTC 98.880 682.430 674.790 2.048 99.412 83.557
C,TTC 98.357 41.990 41.300 0.125 99.410 83.682
CC,CCT 96.451 507.780 489.760 1.486 99.357 85.168
CCC,CCT 96.126 501.490 482.060 1.463 99.301 86.631
C,CCT 94.635 442.690 418.940 1.271 99.230 87.903
T,CCC 93.276 377.620 352.230 1.069 99.154 88.972
T,CTC 90.108 77.030 69.410 0.211 99.130 89.182
T,TCC 85.806 107.300 92.070 0.279 99.082 89.462
C,TCT 75.684 71.230 53.910 0.164 99.026 89.625
T,CCT 66.135 117.970 78.020 0.237 98.897 89.862
T,TTC 43.659 323.963 141.438 0.429 98.305 90.291
TT,TTC 40.714 305.130 124.230 0.377 97.731 90.668
C,CTT 40.404 91.784 37.084 0.113 97.559 90.781
T,TCT 37.706 320.370 120.800 0.367 96.940 91.148
TTT,TTC 36.101 264.650 95.540 0.290 96.425 91.438
T,CTT 30.172 288.380 87.010 0.264 95.819 91.702
C,TTT 21.308 98.130 20.910 0.063 95.588 91.765
TTT,TTT 1.190 227971.488 2713.428 8.235 12.693 100.000

Circuit Grade (%) Recovery (%)


R-C Locked Cycle 98.730 88.242

128
Test 3 Zorba Grade vs. Zorba Recovery
100

90

80

70 Tree Analysis

60
Recovery (%)

50
Rougher
Cleaner Locked
Cycle
40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

129
Aluminum Grade and Recovery:

Sample Al Mass (g) Fluff Mass (g) Total Mass (g) Aluminum Grade (%)
TTT,TTT 832.606 225258.060 227971.488 0.365
TTT,TTC 66.770 169.110 264.650 25.230
TT,TTC 89.860 180.900 305.130 29.450
T,TTC 85.749 182.525 323.963 26.469
T,TCT 75.190 199.570 320.370 23.470
T,TCC 63.300 15.230 107.300 58.993
T,CTT 53.400 201.370 288.380 18.517
T,CTC 41.560 7.620 77.030 53.953
T,CCT 51.330 39.950 117.970 43.511
T,CCC 250.260 25.390 377.620 66.273
C,TTT 14.750 77.220 98.130 15.031
C,TTC 15.730 0.690 41.990 37.461
C,TCT 38.050 17.320 71.230 53.419
C,TCC 410.630 3.760 518.320 79.223
C,CTT 26.600 54.700 91.784 28.981
C,CTC 526.180 7.640 682.430 77.104
C,CCT 366.360 23.750 442.690 82.758
CC,CCT 447.240 18.020 507.780 88.078
CCC,CCT 347.750 19.430 501.490 69.343
CCC,CCC 23072.411 151.506 26494.315 87.084

130
Total Mass Cumulative Cumulative
Sample Al Grade (%) (g) Al Mass (g) Al Yield (%) Grade (%) Al Mass (%)
CC,CCT 88.078 507.780 447.240 1.664 88.078 1.664
CCC,CCC 87.084 26494.315 23072.411 85.849 87.103 87.513
C,CCT 82.758 442.690 366.360 1.363 87.033 88.876
C,TCC 79.223 518.320 410.630 1.528 86.888 90.404
C,CTC 77.104 682.430 526.180 1.958 86.655 92.361
CCC,CCT 69.343 501.490 347.750 1.294 86.357 93.655
T,CCC 66.273 377.620 250.260 0.931 86.100 94.587
T,TCC 58.993 107.300 63.300 0.236 86.002 94.822
T,CTC 53.953 77.030 41.560 0.155 85.919 94.977
C,TCT 53.419 71.230 38.050 0.142 85.841 95.118
T,CCT 43.511 117.970 51.330 0.191 85.674 95.309
C,TTC 37.461 41.990 15.730 0.059 85.607 95.368
TT,TTC 29.450 305.130 89.860 0.334 85.040 95.702
C,CTT 28.981 91.784 26.600 0.099 84.871 95.801
T,TTC 26.469 323.963 85.749 0.319 84.254 96.120
TTT,TTC 25.230 264.650 66.770 0.248 83.748 96.369
T,TCT 23.470 320.370 75.190 0.280 83.130 96.648
T,CTT 18.517 288.380 53.400 0.199 82.539 96.847
C,TTT 15.031 98.130 14.750 0.055 82.330 96.902
TTT,TTT 0.365 227971.488 832.606 3.098 10.353 100.000

Circuit Al Grade (%) Recovery (%)


Rougher Cleaner Locked Cycle 85.791 94.009

131
Test 3 Aluminum Grade vs. Aluminum Recovery
100

90

80

Tree Analysis
70

60
Recovery (%)

50

Rougher
40 Cleaner
Locked Cycle

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

132
Reds Grade and Recovery:

Sample Reds Mass (g) Fluff Mass (g) Total Mass (g) Reds Grade (%)
TTT,TTT 1880.822 225258.060 227971.488 0.825
TTT,TTC 28.770 169.110 264.650 10.871
TT,TTC 34.370 180.900 305.130 11.264
T,TTC 55.689 182.525 323.963 17.190
T,TCT 45.610 199.570 320.370 14.237
T,TCC 28.770 15.230 107.300 26.813
T,CTT 33.610 201.370 288.380 11.655
T,CTC 27.850 7.620 77.030 36.155
T,CCT 26.690 39.950 117.970 22.624
T,CCC 101.970 25.390 377.620 27.003
C,TTT 6.160 77.220 98.130 6.277
C,TTC 25.570 0.690 41.990 60.895
C,TCT 15.860 17.320 71.230 22.266
C,TCC 103.930 3.760 518.320 20.051
C,CTT 10.484 54.700 91.784 11.422
C,CTC 148.610 7.640 682.430 21.777
C,CCT 52.580 23.750 442.690 11.877
CC,CCT 42.520 18.020 507.780 8.374
CCC,CCT 134.310 19.430 501.490 26.782
CCC,CCC 3270.398 151.506 26494.315 12.344

133
Cumulative
Reds Grade Total Mass Reds Mass Reds Yield Cumulative Reds Mass
Sample (%) (g) (g) (%) Grade (%) (%)
C,TTC 60.895 41.990 25.570 0.421 60.895 0.421
T,CTC 36.155 77.030 27.850 0.458 44.883 0.879
T,CCC 27.003 377.620 101.970 1.679 31.288 2.558
T,TCC 26.813 107.300 28.770 0.474 30.493 3.032
CCC,CCT 26.782 501.490 134.310 2.211 28.810 5.243
T,CCT 22.624 117.970 26.690 0.439 28.213 5.682
C,TCT 22.266 71.230 15.860 0.261 27.886 5.943
C,CTC 21.777 682.430 148.610 2.446 25.777 8.390
C,TCC 20.051 518.320 103.930 1.711 24.588 10.100
T,TTC 17.190 323.963 55.689 0.917 23.738 11.017
T,TCT 14.237 320.370 45.610 0.751 22.768 11.768
CCC,CCC 12.344 26494.315 3270.398 53.837 13.448 65.606
C,CCT 11.877 442.690 52.580 0.866 13.425 66.471
T,CTT 11.655 288.380 33.610 0.553 13.408 67.024
C,CTT 11.422 91.784 10.484 0.173 13.402 67.197
TT,TTC 11.264 305.130 34.370 0.566 13.381 67.763
TTT,TTC 10.871 264.650 28.770 0.474 13.360 68.236
CC,CCT 8.374 507.780 42.520 0.700 13.279 68.936
C,TTT 6.277 98.130 6.160 0.101 13.258 69.038
TTT,TTT 0.825 227971.488 1880.822 30.962 2.340 100.000

Circuit Grade (%) Recovery (%)


R-C Locked Cycle 12.109 64.829

134
Test 3 Reds Grade vs. Reds Recovery
100

90

80
Tree Analysis

70

60
Recovery (%)

50
Rougher Cleaner
Locked Cycle
40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

135
Specific Reds Grade and Recovery:

Total Relevant Specific Reds


Sample Reds Mass (g) Fluff Mass (g) Mass (g) Grade (%)
TTT,TTT 1880.822 225258.060 227138.882 0.828
TTT,TTC 28.770 169.110 197.880 14.539
TT,TTC 34.370 180.900 215.270 15.966
T,TTC 55.689 182.525 238.214 23.378
T,TCT 45.610 199.570 245.180 18.603
T,TCC 28.770 15.230 44.000 65.386
T,CTT 33.610 201.370 234.980 14.303
T,CTC 27.850 7.620 35.470 78.517
T,CCT 26.690 39.950 66.640 40.051
T,CCC 101.970 25.390 127.360 80.064
C,TTT 6.160 77.220 83.380 7.388
C,TTC 25.570 0.690 26.260 97.372
C,TCT 15.860 17.320 33.180 47.800
C,TCC 103.930 3.760 107.690 96.508
C,CTT 10.484 54.700 65.184 16.084
C,CTC 148.610 7.640 156.250 95.110
C,CCT 52.580 23.750 76.330 68.885
CC,CCT 42.520 18.020 60.540 70.235
CCC,CCT 134.310 19.430 153.740 87.362
CCC,CCC 3270.398 151.506 3421.904 95.572

136
Specific Total Cumulative Cumulative
Reds Grade Relevant Reds Mass Reds Yield Specific Reds Mass
Sample (%) Mass (g) (g) (%) Grade (%) (%)
C,TTC 97.372 26.260 25.570 0.421 97.372 0.421
C,TCC 96.508 107.690 103.930 1.711 96.678 2.132
CCC,CCC 95.572 3421.904 3270.398 53.837 95.614 55.969
C,CTC 95.110 156.250 148.610 2.446 95.593 58.416
CCC,CCT 87.362 153.740 134.310 2.211 95.266 60.627
T,CCC 80.064 127.360 101.970 1.679 94.781 62.305
T,CTC 78.517 35.470 27.850 0.458 94.638 62.764
CC,CCT 70.235 60.540 42.520 0.700 94.276 63.464
C,CCT 68.885 76.330 52.580 0.866 93.811 64.329
T,TCC 65.386 44.000 28.770 0.474 93.514 64.803
C,TCT 47.800 33.180 15.860 0.261 93.156 65.064
T,CCT 40.051 66.640 26.690 0.439 92.335 65.503
T,TTC 23.378 238.214 55.689 0.917 88.723 66.420
T,TCT 18.603 245.180 45.610 0.751 85.136 67.171
C,CTT 16.084 65.184 10.484 0.173 84.209 67.344
TT,TTC 15.966 215.270 34.370 0.566 81.314 67.909
TTT,TTC 14.539 197.880 28.770 0.474 78.807 68.383
T,CTT 14.303 234.980 33.610 0.553 76.054 68.936
C,TTT 7.388 83.380 6.160 0.101 75.030 69.038
TTT,TTT 0.828 227138.882 1880.822 30.962 2.610 100.000

Circuit Specific Grade (%) Recovery (%)


R-C Locked Cycle 91.061 62.727

137
Specific Reds Grade vs. Reds Recovery
100

90

80

70 Tree
Analysis

60
Recovery (%)

50
Rougher
Cleaner
40 Locked Cycle

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

138
Fluff Rejection and Zorba Recovery:

Sample Zorba Mass (g) Fluff Mass (g) Total Mass (g) Zorba Grade (%)
TTT,TTT 2713.428 225258.060 227971.488 1.190
TTT,TTC 95.540 169.110 264.650 36.101
TT,TTC 124.230 180.900 305.130 40.714
T,TTC 141.438 182.525 323.963 43.659
T,TCT 120.800 199.570 320.370 37.706
T,TCC 92.070 15.230 107.300 85.806
T,CTT 87.010 201.370 288.380 30.172
T,CTC 69.410 7.620 77.030 90.108
T,CCT 78.020 39.950 117.970 66.135
T,CCC 352.230 25.390 377.620 93.276
C,TTT 20.910 77.220 98.130 21.308
C,TTC 41.300 0.690 41.990 98.357
C,TCT 53.910 17.320 71.230 75.684
C,TCC 514.560 3.760 518.320 99.275
C,CTT 37.084 54.700 91.784 40.404
C,CTC 674.790 7.640 682.430 98.880
C,CCT 418.940 23.750 442.690 94.635
CC,CCT 489.760 18.020 507.780 96.451
CCC,CCT 482.060 19.430 501.490 96.126
CCC,CCC 26342.809 151.506 26494.315 99.428

139
Cumulative Cumulative Separation
Zorba Grade Fluff Yield Zorba Yield Fluff Yield to Zorba Mass Efficiency
Sample (%) (%) (%) T (%) (%) (%)
CCC,CCC 99.428 0.067 79.947 99.933 79.947 79.880
C,TCC 99.275 0.002 1.562 99.931 81.509 81.440
C,CTC 98.880 0.003 2.048 99.928 83.557 83.485
C,TTC 98.357 0.000 0.125 99.928 83.682 83.610
CC,CCT 96.451 0.008 1.486 99.920 85.168 85.088
CCC,CCT 96.126 0.009 1.463 99.911 86.631 86.543
C,CCT 94.635 0.010 1.271 99.901 87.903 87.804
T,CCC 93.276 0.011 1.069 99.890 88.972 88.861
T,CTC 90.108 0.003 0.211 99.886 89.182 89.069
T,TCC 85.806 0.007 0.279 99.880 89.462 89.341
C,TCT 75.684 0.008 0.164 99.872 89.625 89.497
T,CCT 66.135 0.018 0.237 99.854 89.862 89.716
T,TTC 43.659 0.081 0.429 99.774 90.291 90.065
TT,TTC 40.714 0.080 0.377 99.694 90.668 90.362
C,CTT 40.404 0.024 0.113 99.670 90.781 90.451
T,TCT 37.706 0.088 0.367 99.582 91.148 90.729
TTT,TTC 36.101 0.075 0.290 99.507 91.438 90.945
T,CTT 30.172 0.089 0.264 99.418 91.702 91.120
C,TTT 21.308 0.034 0.063 99.384 91.765 91.149
TTT,TTT 1.190 99.384 8.235 0.000 100.000 0.000

Fluff Rejection Zorba Recovery Separation Efficiency Theoretical Efficiency


Circuit (%) (%) (%) (%)
R-C Locked
Cycle 99.835 88.242 88.077 96.630

140
Test 3: Fluff Rejection vs. Zorba Recovery
100

90
Tree
Analysis
80

R-C
70 Locked
Cycle

SE = 80
60
Recovery (%)

50
SE = 90

40
SE = 88

30

SE =
20 91.2

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

141
Rejection Efficiency, Recovery Efficiency, and Spline Fitting

Cumulative
Fluff Zorba Mass
Sample Rejection (x) (y) s dy dx m
CCC,CCC 99.933 79.947 -941.3513 1.5616 -0.0017
C,TCC 99.931 81.509 -607.5456 2.0479 -0.0034 -774.4484
C,CTC 99.928 83.557 -411.7224 0.1253 -0.0003 -509.6340
C,TTC 99.928 83.682 -186.9529 1.4864 -0.0080 -299.3376
CC,CCT 99.920 85.168 -170.6600 1.4630 -0.0086 -178.8065
CCC,CCT 99.911 86.631 -121.3366 1.2714 -0.0105 -145.9983
C,CCT 99.901 87.903 -95.4261 1.0690 -0.0112 -108.3813
T,CCC 99.890 88.972 -62.6572 0.2107 -0.0034 -79.0416
T,CTC 99.886 89.182 -41.5836 0.2794 -0.0067 -52.1204
T,TCC 99.880 89.462 -21.4104 0.1636 -0.0076 -31.4970
C,TCT 99.872 89.625 -13.4336 0.2368 -0.0176 -17.4220
T,CCT 99.854 89.862 -5.3302 0.4292 -0.0805 -9.3819
T,TTC 99.774 90.291 -4.7238 0.3770 -0.0798 -5.0270
TT,TTC 99.694 90.668 -4.6634 0.1125 -0.0241 -4.6936
C,CTT 99.670 90.781 -4.1637 0.3666 -0.0881 -4.4135
T,TCT 99.582 91.148 -3.8861 0.2900 -0.0746 -4.0249
TTT,TTC 99.507 91.438 -2.9722 0.2641 -0.0888 -3.4292
T,CTT 99.418 91.702 -1.8626 0.0635 -0.0341 -2.4174
C,TTT 99.384 91.765 -0.0829 8.2349 -99.3842 -0.9727
TTT,TTT 0.000 100.000

142
Sample a b c d
CCC,CCC 79.9471 0.0000 1235507.4796 402707361.5218
C,TCC 81.5087 -774.4484 -69982.0856 -6072026.6858
C,CTC 83.5566 -509.6340 -274081.3330 156168376.8927
C,TTC 83.6820 -299.3376 -27246.6471 -1649088.5399
CC,CCT 85.1683 -178.8065 976.2473 224733.3144
CCC,CCT 86.6313 -145.9983 -3470.7341 -106616.7119
C,CCT 87.9028 -108.3813 -850.3759 27327.3842
T,CCC 88.9717 -79.0416 -6612.8669 -517368.8261
T,CTC 89.1824 -52.1204 -1635.0913 -9971.0015
T,TCC 89.4618 -31.4970 -2117.9775 -104431.2177
C,TCT 89.6254 -17.4220 -222.6892 203.6872
T,CCT 89.8622 -9.3819 -96.8597 -578.0079
T,TTC 90.2914 -5.0270 -7.2199 -42.8598
TT,TTC 90.6685 -4.6936 7.8509 377.1598
C,CTT 90.7810 -4.4135 -4.0998 -14.3321
T,TCT 91.1476 -4.0249 2.4053 57.1629
TTT,TTC 91.4376 -3.4292 -4.0426 12.3910
T,CTT 91.7016 -2.4174 -6.4479 288.7009
C,TTT 91.7651 -0.9727 -0.0171 -0.0001
TTT,TTT

Fluff Zorba Efficient Efficient Rejection Recovery


Rejection Recovery Rejection Recovery Efficiency Efficiency
Test (%) (%) (%) (%) (%) (%)
R-C Locked
Cycle 99.835 88.242 99.896 89.979 99.939 98.070

143
Test 3: Fluff Rejection vs. Zorba Recovery
100

90

Tree Analysis
80

70
Rougher Cleaner
Locked Cycle
60
Recovery (%)

50 Efficent
Rejection

40

Efficent
30 Recovery

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

144
APPENDIX F: TREE ANALYSIS FOR TEST 4

145
Test 4 Sample Diagram:

146
Zorba Grade and Recovery:

Sample Reds Mass (g) Fluff Mass (g) Zorba Metals (g) Total Mass (g)
TTT,TTT 2720.57 227427.28 3917.43 231344.71
TTT,TTC 35.61 191.09 80.49 271.58
TT,TTC 41.81 186.80 88.77 275.57
T,TTC 49.56 221.61 109.80 331.41
T,TCT 45.92 188.58 100.56 289.14
T,TCC 17.40 9.68 44.14 53.82
T,CCT 43.80 226.32 92.60 318.92
T,CTC 17.25 9.52 35.60 45.12
TCC 29.98 16.75 72.01 88.76
C,TTT 63.68 964.46 150.37 1114.83
C,TTC 17.87 12.80 75.27 88.07
CTC 51.36 16.14 292.05 308.19
C,CTT 27.20 116.62 64.84 181.46
C,CTC 67.00 7.50 274.92 282.42
C,CCT 62.13 54.70 560.69 615.39
CC,CCT 90.45 161.83 373.12 534.95
CCC,CCT 44.13 29.77 495.63 525.40
CCC,CCC 3032.72 124.28 26010.78 26135.06

147
Cumulative
Zorba Grade Total Mass Zorba Mass Zorba Yield Cumulative Zorba Mass
Sample (%) (g) (g) (%) Grade (%) (%)
CCC,CCC 99.524 26135.064 26010.780 79.207 99.524 79.207
C,CTC 97.344 282.420 274.920 0.837 99.501 80.044
CTC 94.763 308.190 292.050 0.889 99.447 80.933
CCC,CCT 94.334 525.400 495.630 1.509 99.348 82.443
C,CCT 91.111 615.390 560.690 1.707 99.166 84.150
C,TTC 85.466 88.070 75.270 0.229 99.123 84.379
T,TCC 82.014 53.820 44.140 0.134 99.090 84.514
TCC 81.129 88.760 72.010 0.219 99.033 84.733
T,CTC 78.901 45.120 35.600 0.108 99.001 84.841
CC,CCT 69.749 534.950 373.120 1.136 98.455 85.977
C,CTT 35.732 181.460 64.840 0.197 98.061 86.175
T,TCT 34.779 289.140 100.560 0.306 97.433 86.481
T,TTC 33.131 331.410 109.800 0.334 96.710 86.816
TT,TTC 32.213 275.570 88.770 0.270 96.113 87.086
TTT,TTC 29.638 271.580 80.490 0.245 95.512 87.331
T,CCT 29.035 318.920 92.600 0.282 94.813 87.613
C,TTT 13.488 1114.830 150.370 0.458 91.931 88.071
TTT,TTT 1.693 231344.709 3917.434 11.929 12.496 100.000

Circuit Grade (%) Recovery (%)


R-S Locked Cycle 95.004 86.171

148
Test 4 Zorba Grade vs. Zorba Recovery
100

90

80

Tree Analysis
70

60
Recovery (%)

50

Rougher
40 Scavenger
Locked Cycle

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

149
Aluminum Grade and Recovery:

Aluminum Grade
Sample Al Mass (g) Fluff Mass (g) Total Mass (g) (%)
TTT,TTT 1196.866 227427.276 231344.709 0.517
TTT,TTC 44.880 191.090 271.580 16.526
TT,TTC 46.960 186.800 275.570 17.041
T,TTC 60.240 221.610 331.410 18.177
T,TCT 54.640 188.580 289.140 18.897
T,TCC 26.740 9.680 53.820 49.684
T,CCT 48.800 226.320 318.920 15.302
T,CTC 18.350 9.520 45.120 40.669
TCC 42.030 16.750 88.760 47.352
C,TTT 86.690 964.460 1114.830 7.776
C,TTC 57.400 12.800 88.070 65.175
CTC 240.690 16.140 308.190 78.098
C,CTT 37.640 116.620 181.460 20.743
C,CTC 207.920 7.500 282.420 73.621
C,CCT 498.560 54.700 615.390 81.015
CC,CCT 282.670 161.830 534.950 52.840
CCC,CCT 451.500 29.770 525.400 85.935
CCC,CCC 22978.064 124.284 26135.064 87.920

150
Total Mass Cumulative Cumulative
Sample Al Grade (%) (g) Al Mass (g) Al Yield (%) Grade (%) Al Mass (%)
CCC,CCC 87.920 26135.064 22978.064 87.102 87.920 87.102
CCC,CCT 85.935 525.400 451.500 1.711 87.881 88.813
C,CCT 81.015 615.390 498.560 1.890 87.726 90.703
CTC 78.098 308.190 240.690 0.912 87.619 91.616
C,CTC 73.621 282.420 207.920 0.788 87.477 92.404
C,TTC 65.175 88.070 57.400 0.218 87.407 92.621
CC,CCT 52.840 534.950 282.670 1.072 86.758 93.693
T,TCC 49.684 53.820 26.740 0.101 86.688 93.794
TCC 47.352 88.760 42.030 0.159 86.566 93.954
T,CTC 40.669 45.120 18.350 0.070 86.494 94.023
C,CTT 20.743 181.460 37.640 0.143 86.080 94.166
T,TCT 18.897 289.140 54.640 0.207 85.414 94.373
T,TTC 18.177 331.410 60.240 0.228 84.658 94.601
TT,TTC 17.041 275.570 46.960 0.178 84.032 94.779
TTT,TTC 16.526 271.580 44.880 0.170 83.421 94.949
T,CCT 15.302 318.920 48.800 0.185 82.705 95.134
C,TTT 7.776 1114.830 86.690 0.329 80.050 95.463
TTT,TTT 0.517 231344.709 1196.866 4.537 10.038 100.000

Circuit Al Grade (%) Recovery (%)


R-S Locked Cycle 83.399 94.164

151
Test 4 Aluminum Grade vs. Aluminum Recovery
100

90

80

Tree Analysis
70

60
Recovery (%)

Rougher Scavenger
50 Locked Cycle

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

152
Reds Grade and Recovery:

Sample Reds Mass (g) Fluff Mass (g) Total Mass (g) Reds Grade (%)
TTT,TTT 2720.568 227427.276 231344.709 1.176
TTT,TTC 35.610 191.090 271.580 13.112
TT,TTC 41.810 186.800 275.570 15.172
T,TTC 49.560 221.610 331.410 14.954
T,TCT 45.920 188.580 289.140 15.882
T,TCC 17.400 9.680 53.820 32.330
T,CCT 43.800 226.320 318.920 13.734
T,CTC 17.250 9.520 45.120 38.231
TCC 29.980 16.750 88.760 33.776
C,TTT 63.680 964.460 1114.830 5.712
C,TTC 17.870 12.800 88.070 20.291
CTC 51.360 16.140 308.190 16.665
C,CTT 27.200 116.620 181.460 14.990
C,CTC 67.000 7.500 282.420 23.724
C,CCT 62.130 54.700 615.390 10.096
CC,CCT 90.450 161.830 534.950 16.908
CCC,CCT 44.130 29.770 525.400 8.399
CCC,CCC 3032.716 124.284 26135.064 11.604

153
Cumulative
Reds Grade Total Mass Reds Mass Reds Yield Cumulative Reds Mass
Sample (%) (g) (g) (%) Grade (%) (%)
T,CTC 38.231 45.120 17.250 0.267 38.231 0.267
TCC 33.776 88.760 29.980 0.464 35.278 0.731
T,TCC 32.330 53.820 17.400 0.269 34.433 1.001
C,CTC 23.724 282.420 67.000 1.037 27.999 2.038
C,TTC 20.291 88.070 17.870 0.277 26.783 2.315
CC,CCT 16.908 534.950 90.450 1.400 21.951 3.715
CTC 16.665 308.190 51.360 0.795 20.788 4.511
T,TCT 15.882 289.140 45.920 0.711 19.949 5.222
TT,TTC 15.172 275.570 41.810 0.647 19.279 5.869
C,CTT 14.990 181.460 27.200 0.421 18.917 6.290
T,TTC 14.954 331.410 49.560 0.767 18.387 7.057
T,CCT 13.734 318.920 43.800 0.678 17.857 7.736
TTT,TTC 13.112 271.580 35.610 0.551 17.437 8.287
CCC,CCC 11.604 26135.064 3032.716 46.957 12.217 55.244
C,CCT 10.096 615.390 62.130 0.962 12.173 56.206
CCC,CCT 8.399 525.400 44.130 0.683 12.108 56.890
C,TTT 5.712 1114.830 63.680 0.986 11.881 57.876
TTT,TTT 1.176 231344.709 2720.568 42.124 2.458 100.000

Circuit Grade (%) Recovery (%)


R-S Locked Cycle 11.605 53.520

154
Test 4 Reds Grade vs. Reds Recovery
100

90

80
Tree Analysis
70

60
Recovery (%)

Rougher
50 Scavenger
Locked Cycle

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

155
Specific Reds Grade and Recovery:

Total Relevant Specific Reds


Sample Reds Mass (g) Fluff Mass (g) Mass (g) Grade (%)
TTT,TTT 2720.568 227427.276 230147.844 1.182
TTT,TTC 35.610 191.090 226.700 15.708
TT,TTC 41.810 186.800 228.610 18.289
T,TTC 49.560 221.610 271.170 18.276
T,TCT 45.920 188.580 234.500 19.582
T,TCC 17.400 9.680 27.080 64.254
T,CCT 43.800 226.320 270.120 16.215
T,CTC 17.250 9.520 26.770 64.438
TCC 29.980 16.750 46.730 64.156
C,TTT 63.680 964.460 1028.140 6.194
C,TTC 17.870 12.800 30.670 58.265
CTC 51.360 16.140 67.500 76.089
C,CTT 27.200 116.620 143.820 18.913
C,CTC 67.000 7.500 74.500 89.933
C,CCT 62.130 54.700 116.830 53.180
CC,CCT 90.450 161.830 252.280 35.853
CCC,CCT 44.130 29.770 73.900 59.716
CCC,CCC 3032.716 124.284 3157.000 96.063

156
Specific Total Cumulative Cumulative
Reds Grade Relevant Reds Mass Reds Yield Specific Reds Mass
Sample (%) Mass (g) (g) (%) Grade (%) (%)
CCC,CCC 96.063 3157.000 3032.716 46.957 96.063 46.957
C,CTC 89.933 74.500 67.000 1.037 95.922 47.995
CTC 76.089 67.500 51.360 0.795 95.516 48.790
T,CTC 64.438 26.770 17.250 0.267 95.266 49.057
T,TCC 64.254 27.080 17.400 0.269 95.015 49.327
TCC 64.156 46.730 29.980 0.464 94.591 49.791
CCC,CCT 59.716 73.900 44.130 0.683 93.849 50.474
C,TTC 58.265 30.670 17.870 0.277 93.538 50.751
C,CCT 53.180 116.830 62.130 0.962 92.236 51.713
CC,CCT 35.853 252.280 90.450 1.400 88.563 53.113
T,TCT 19.582 234.500 45.920 0.711 84.625 53.824
C,CTT 18.913 143.820 27.200 0.421 82.402 54.245
TT,TTC 18.289 228.610 41.810 0.647 79.131 54.893
T,TTC 18.276 271.170 49.560 0.767 75.658 55.660
T,CCT 16.215 270.120 43.800 0.678 72.460 56.338
TTT,TTC 15.708 226.700 35.610 0.551 70.009 56.890
C,TTT 6.194 1028.140 63.680 0.986 59.555 57.876
TTT,TTT 1.182 230147.844 2720.568 42.124 2.732 100.000

Circuit Specific Grade (%) Recovery (%)


R-S Locked Cycle 69.905 53.520

157
Specific Reds Grade vs. Reds Recovery
100

90

80
Tree Analysis
70

60
Recovery (%)

Rougher
50 Scavenger
Locked Cycle

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Grade (%)

158
Fluff Rejection and Zorba Recovery:

Sample Zorba Mass (g) Fluff Mass (g) Total Mass (g) Zorba Grade (%)
TTT,TTT 3917.434 227427.276 231344.709 1.693
TTT,TTC 80.490 191.090 271.580 29.638
TT,TTC 88.770 186.800 275.570 32.213
T,TTC 109.800 221.610 331.410 33.131
T,TCT 100.560 188.580 289.140 34.779
T,TCC 44.140 9.680 53.820 82.014
T,CCT 92.600 226.320 318.920 29.035
T,CTC 35.600 9.520 45.120 78.901
TCC 72.010 16.750 88.760 81.129
C,TTT 150.370 964.460 1114.830 13.488
C,TTC 75.270 12.800 88.070 85.466
CTC 292.050 16.140 308.190 94.763
C,CTT 64.840 116.620 181.460 35.732
C,CTC 274.920 7.500 282.420 97.344
C,CCT 560.690 54.700 615.390 91.111
CC,CCT 373.120 161.830 534.950 69.749
CCC,CCT 495.630 29.770 525.400 94.334
CCC,CCC 26010.780 124.284 26135.064 99.524

159
Cumulative Cumulative Separation
Zorba Grade Fluff Yield Zorba Yield Fluff Yield to Zorba Mass Efficiency
Sample (%) (%) (%) T (%) (%) (%)
CCC,CCC 99.524 0.054 79.207 99.946 79.207 79.153
C,CTC 97.344 0.003 0.837 99.943 80.044 79.987
CTC 94.763 0.007 0.889 99.936 80.933 80.869
CCC,CCT 94.334 0.013 1.509 99.923 82.443 82.365
C,CCT 91.111 0.024 1.707 99.899 84.150 84.049
C,TTC 85.466 0.006 0.229 99.893 84.379 84.273
T,TCC 82.014 0.004 0.134 99.889 84.514 84.403
TCC 81.129 0.007 0.219 99.882 84.733 84.615
T,CTC 78.901 0.004 0.108 99.878 84.841 84.719
CC,CCT 69.749 0.070 1.136 99.807 85.977 85.785
C,CTT 35.732 0.051 0.197 99.757 86.175 85.932
T,TCT 34.779 0.082 0.306 99.675 86.481 86.156
T,TTC 33.131 0.096 0.334 99.578 86.816 86.394
TT,TTC 32.213 0.081 0.270 99.497 87.086 86.583
TTT,TTC 29.638 0.083 0.245 99.414 87.331 86.745
T,CCT 29.035 0.098 0.282 99.316 87.613 86.928
C,TTT 13.488 0.419 0.458 98.896 88.071 86.967
TTT,TTT 1.693 98.896 11.929 0.000 100.000 0.000

Zorba Recovery Separation Theoretical


Circuit Fluff Rejection (%) (%) Efficiency (%) Efficiency (%)
R-S Locked Cycle 99.353 86.171 85.524 98.340

160
Test 4: Fluff Rejection vs. Zorba Recovery
100
Tree
Analysis
90

R-S
80 Locked
Cycle

70 SE = 50

60
SE = 80
Recovery (%)

50
SE = 90

40

SE = 87
30

20 SE = 85.5

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

161
Rejection Efficiency, Recovery Efficiency, and Spline Fitting

Cumulative
Fluff Zorba Mass
Sample Rejection (x) (y) s dy dx m
CCC,CCC 99.946 79.207 -256.6949 0.8372 -0.0033
C,CTC 99.943 80.044 -126.7144 0.8893 -0.0070 -191.7047
CTC 99.936 80.933 -116.5872 1.5093 -0.0129 -121.6508
CCC,CCT 99.923 82.443 -71.7807 1.7074 -0.0238 -94.1840
C,CCT 99.899 84.150 -41.1798 0.2292 -0.0056 -56.4802
C,TTC 99.893 84.379 -31.9322 0.1344 -0.0042 -36.5560
T,TCC 99.889 84.514 -30.1058 0.2193 -0.0073 -31.0190
TCC 99.882 84.733 -26.1870 0.1084 -0.0041 -28.1464
T,CTC 99.878 84.841 -16.1459 1.1362 -0.0704 -21.1664
CC,CCT 99.807 85.977 -3.8935 0.1974 -0.0507 -10.0197
C,CTT 99.757 86.175 -3.7342 0.3062 -0.0820 -3.8139
T,TCT 99.675 86.481 -3.4696 0.3344 -0.0964 -3.6019
T,TTC 99.578 86.816 -3.3278 0.2703 -0.0812 -3.3987
TT,TTC 99.497 87.086 -2.9497 0.2451 -0.0831 -3.1388
TTT,TTC 99.414 87.331 -2.8652 0.2820 -0.0984 -2.9075
T,CCT 99.316 87.613 -1.0918 0.4579 -0.4194 -1.9785
C,TTT 98.896 88.071 -0.1206 11.9292 -98.8962 -0.6062
TTT,TTT 0.000 100.000

162
Sample a b c d
CCC,CCC 79.2068 0.0000 177343.4837 30243692.4041
C,CTC 80.0440 -191.7047 -17798.4039 -1216576.8494
CTC 80.9333 -121.6508 948.2985 103468.8932
CCC,CCT 82.4426 -94.1840 -1240.4726 -12553.9864
C,CCT 84.1500 -56.4802 -4667.0692 -344621.3257
C,TTC 84.3792 -36.5560 -1979.9757 -209419.1919
T,TCC 84.5136 -31.0190 18.2591 19720.2921
TCC 84.7329 -28.1464 266.1306 178621.6395
T,CTC 84.8413 -21.1664 -55.6322 223.2647
CC,CCT 85.9775 -10.0197 -240.0368 -2351.1867
C,CTT 86.1749 -3.8139 -0.3291 7.8302
T,TCT 86.4812 -3.6019 -2.0099 -6.6105
T,TTC 86.8155 -3.3987 0.5818 17.9087
TT,TTC 87.0858 -3.1388 -4.0426 -21.2672
TTT,TTC 87.3309 -2.9075 8.1518 87.1909
T,CCT 87.6129 -1.9785 -3.0707 -2.2805
C,TTT 88.0708 -0.6062 -0.0086 0.0000
TTT,TTT

Fluff Zorba Efficient Efficient Rejection Recovery


Rejection Recovery Rejection Recovery Efficiency Efficiency
Test (%) (%) (%) (%) (%) (%)
R-S Locked
Cycle 99.353 86.171 99.758 87.527 99.594 98.451

163
Test 4: Fluff Rejection vs. Zorba Recovery
100

90

80
Tree Analysis

70

60 Rougher
Scavenger
Recovery (%)

Locked Cycle
50

Efficent
40 Rejection

30
Efficent
Recovery
20

10

0
0 10 20 30 40 50 60 70 80 90 100
Rejection (%)

164
165